chore: bump version to 1.0.6

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
fix: raise speed minimum from 0.1 to 0.3 to prevent VRAM explosion
2026-04-10 22:17:21 +02:00 · 2026-04-10 22:12:04 +02:00 · 2026-04-10 12:14:13 +02:00 · 2026-04-10 12:13:19 +02:00 · 2026-04-10 12:08:44 +02:00 · 2026-04-10 00:07:00 +02:00
18 changed files with 1591 additions and 155 deletions
@@ -0,0 +1,19 @@
 name: Publish to ComfyUI Registry
 on:
  workflow_dispatch:
  push:
    branches:
      - master
    paths:
      - "pyproject.toml"
 jobs:
  publish:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - name: Publish to ComfyUI Registry
        uses: Comfy-Org/publish-node-action@main
        with:
          personal_access_token: ${{ secrets.REGISTRY_ACCESS_TOKEN }}
@@ -0,0 +1,674 @@
                    GNU GENERAL PUBLIC LICENSE
                       Version 3, 29 June 2007
 Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
 Everyone is permitted to copy and distribute verbatim copies
 of this license document, but changing it is not allowed.
                            Preamble
  The GNU General Public License is a free, copyleft license for
 software and other kinds of works.
  The licenses for most software and other practical works are designed
 to take away your freedom to share and change the works.  By contrast,
 the GNU General Public License is intended to guarantee your freedom to
 share and change all versions of a program--to make sure it remains free
 software for all its users.  We, the Free Software Foundation, use the
 GNU General Public License for most of our software; it applies also to
 any other work released this way by its authors.  You can apply it to
 your programs, too.
  When we speak of free software, we are referring to freedom, not
 price.  Our General Public Licenses are designed to make sure that you
 have the freedom to distribute copies of free software (and charge for
 them if you wish), that you receive source code or can get it if you
 want it, that you can change the software or use pieces of it in new
 free programs, and that you know you can do these things.
  To protect your rights, we need to prevent others from denying you
 these rights or asking you to surrender the rights.  Therefore, you have
 certain responsibilities if you distribute copies of the software, or if
 you modify it: responsibilities to respect the freedom of others.
  For example, if you distribute copies of such a program, whether
 gratis or for a fee, you must pass on to the recipients the same
 freedoms that you received.  You must make sure that they, too, receive
 or can get the source code.  And you must show them these terms so they
 know their rights.
  Developers that use the GNU GPL protect your rights with two steps:
 (1) assert copyright on the software, and (2) offer you this License
 giving you legal permission to copy, distribute and/or modify it.
  For the developers' and authors' protection, the GPL clearly explains
 that there is no warranty for this free software.  For both users' and
 authors' sake, the GPL requires that modified versions be marked as
 changed, so that their problems will not be attributed erroneously to
 authors of previous versions.
  Some devices are designed to deny users access to install or run
 modified versions of the software inside them, although the manufacturer
 can do so.  This is fundamentally incompatible with the aim of
 protecting users' freedom to change the software.  The systematic
 pattern of such abuse occurs in the area of products for individuals to
 use, which is precisely where it is most unacceptable.  Therefore, we
 have designed this version of the GPL to prohibit the practice for those
 products.  If such problems arise substantially in other domains, we
 stand ready to extend this provision to those domains in future versions
 of the GPL, as needed to protect the freedom of users.
  Finally, every program is threatened constantly by software patents.
 States should not allow patents to restrict development and use of
 software on general-purpose computers, but in those that do, we wish to
 avoid the special danger that patents applied to a free program could
 make it effectively proprietary.  To prevent this, the GPL assures that
 patents cannot be used to render the program non-free.
  The precise terms and conditions for copying, distribution and
 modification follow.
                       TERMS AND CONDITIONS
  0. Definitions.
  "This License" refers to version 3 of the GNU General Public License.
  "Copyright" also means copyright-like laws that apply to other kinds of
 works, such as semiconductor masks.
  "The Program" refers to any copyrightable work licensed under this
 License.  Each licensee is addressed as "you".  "Licensees" and
 "recipients" may be individuals or organizations.
  To "modify" a work means to copy from or adapt all or part of the work
 in a fashion requiring copyright permission, other than the making of an
 exact copy.  The resulting work is called a "modified version" of the
 earlier work or a work "based on" the earlier work.
  A "covered work" means either the unmodified Program or a work based
 on the Program.
  To "propagate" a work means to do anything with it that, without
 permission, would make you directly or secondarily liable for
 infringement under applicable copyright law, except executing it on a
 computer or modifying a private copy.  Propagation includes copying,
 distribution (with or without modification), making available to the
 public, and in some countries other activities as well.
  To "convey" a work means any kind of propagation that enables other
 parties to make or receive copies.  Mere interaction with a user through
 a computer network, with no transfer of a copy, is not conveying.
  An interactive user interface displays "Appropriate Legal Notices"
 to the extent that it includes a convenient and prominently visible
 feature that (1) displays an appropriate copyright notice, and (2)
 tells the user that there is no warranty for the work (except to the
 extent that warranties are provided), that licensees may convey the
 work under this License, and how to view a copy of this License.  If
 the interface presents a list of user commands or options, such as a
 menu, a prominent item in the list meets this criterion.
  1. Source Code.
  The "source code" for a work means the preferred form of the work
 for making modifications to it.  "Object code" means any non-source
 form of a work.
  A "Standard Interface" means an interface that either is an official
 standard defined by a recognized standards body, or, in the case of
 interfaces specified for a particular programming language, one that
 is widely used among developers working in that language.
  The "System Libraries" of an executable work include anything, other
 than the work as a whole, that (a) is included in the normal form of
 packaging a Major Component, but which is not part of that Major
 Component, and (b) serves only to enable use of the work with that
 Major Component, or to implement a Standard Interface for which an
 implementation is available to the public in source code form.  A
 "Major Component", in this context, means a major essential component
 (kernel, window system, and so on) of the specific operating system
 (if any) on which the executable work runs, or a compiler used to
 produce the work, or an object code interpreter used to run it.
  The "Corresponding Source" for a work in object code form means all
 the source code needed to generate, install, and (for an executable
 work) run the object code and to modify the work, including scripts to
 control those activities.  However, it does not include the work's
 System Libraries, or general-purpose tools or generally available free
 programs which are used unmodified in performing those activities but
 which are not part of the work.  For example, Corresponding Source
 includes interface definition files associated with source files for
 the work, and the source code for shared libraries and dynamically
 linked subprograms that the work is specifically designed to require,
 such as by intimate data communication or control flow between those
 subprograms and other parts of the work.
  The Corresponding Source need not include anything that users
 can regenerate automatically from other parts of the Corresponding
 Source.
  The Corresponding Source for a work in source code form is that
 same work.
  2. Basic Permissions.
  All rights granted under this License are granted for the term of
 copyright on the Program, and are irrevocable provided the stated
 conditions are met.  This License explicitly affirms your unlimited
 permission to run the unmodified Program.  The output from running a
 covered work is covered by this License only if the output, given its
 content, constitutes a covered work.  This License acknowledges your
 rights of fair use or other equivalent, as provided by copyright law.
  You may make, run and propagate covered works that you do not
 convey, without conditions so long as your license otherwise remains
 in force.  You may convey covered works to others for the sole purpose
 of having them make modifications exclusively for you, or provide you
 with facilities for running those works, provided that you comply with
 the terms of this License in conveying all material for which you do
 not control copyright.  Those thus making or running the covered works
 for you must do so exclusively on your behalf, under your direction
 and control, on terms that prohibit them from making any copies of
 your copyrighted material outside their relationship with you.
  Conveying under any other circumstances is permitted solely under
 the conditions stated below.  Sublicensing is not allowed; section 10
 makes it unnecessary.
  3. Protecting Users' Legal Rights From Anti-Circumvention Law.
  No covered work shall be deemed part of an effective technological
 measure under any applicable law fulfilling obligations under article
 11 of the WIPO copyright treaty adopted on 20 December 1996, or
 similar laws prohibiting or restricting circumvention of such
 measures.
  When you convey a covered work, you waive any legal power to forbid
 circumvention of technological measures to the extent such circumvention
 is effected by exercising rights under this License with respect to
 the covered work, and you disclaim any intention to limit operation or
 modification of the work as a means of enforcing, against the work's
 users, your or third parties' legal rights to forbid circumvention of
 technological measures.
  4. Conveying Verbatim Copies.
  You may convey verbatim copies of the Program's source code as you
 receive it, in any medium, provided that you conspicuously and
 appropriately publish on each copy an appropriate copyright notice;
 keep intact all notices stating that this License and any
 non-permissive terms added in accord with section 7 apply to the code;
 keep intact all notices of the absence of any warranty; and give all
 recipients a copy of this License along with the Program.
  You may charge any price or no price for each copy that you convey,
 and you may offer support or warranty protection for a fee.
  5. Conveying Modified Source Versions.
  You may convey a work based on the Program, or the modifications to
 produce it from the Program, in the form of source code under the
 terms of section 4, provided that you also meet all of these conditions:
    a) The work must carry prominent notices stating that you modified
    it, and giving a relevant date.
    b) The work must carry prominent notices stating that it is
    released under this License and any conditions added under section
    7.  This requirement modifies the requirement in section 4 to
    "keep intact all notices".
    c) You must license the entire work, as a whole, under this
    License to anyone who comes into possession of a copy.  This
    License will therefore apply, along with any applicable section 7
    additional terms, to the whole of the work, and all its parts,
    regardless of how they are packaged.  This License gives no
    permission to license the work in any other way, but it does not
    invalidate such permission if you have separately received it.
    d) If the work has interactive user interfaces, each must display
    Appropriate Legal Notices; however, if the Program has interactive
    interfaces that do not display Appropriate Legal Notices, your
    work need not make them do so.
  A compilation of a covered work with other separate and independent
 works, which are not by their nature extensions of the covered work,
 and which are not combined with it such as to form a larger program,
 in or on a volume of a storage or distribution medium, is called an
 "aggregate" if the compilation and its resulting copyright are not
 used to limit the access or legal rights of the compilation's users
 beyond what the individual works permit.  Inclusion of a covered work
 in an aggregate does not cause this License to apply to the other
 parts of the aggregate.
  6. Conveying Non-Source Forms.
  You may convey a covered work in object code form under the terms
 of sections 4 and 5, provided that you also convey the
 machine-readable Corresponding Source under the terms of this License,
 in one of these ways:
    a) Convey the object code in, or embodied in, a physical product
    (including a physical distribution medium), accompanied by the
    Corresponding Source fixed on a durable physical medium
    customarily used for software interchange.
    b) Convey the object code in, or embodied in, a physical product
    (including a physical distribution medium), accompanied by a
    written offer, valid for at least three years and valid for as
    long as you offer spare parts or customer support for that product
    model, to give anyone who possesses the object code either (1) a
    copy of the Corresponding Source for all the software in the
    product that is covered by this License, on a durable physical
    medium customarily used for software interchange, for a price no
    more than your reasonable cost of physically performing this
    conveying of source, or (2) access to copy the
    Corresponding Source from a network server at no charge.
    c) Convey individual copies of the object code with a copy of the
    written offer to provide the Corresponding Source.  This
    alternative is allowed only occasionally and noncommercially, and
    only if you received the object code with such an offer, in accord
    with subsection 6b.
    d) Convey the object code by offering access from a designated
    place (gratis or for a charge), and offer equivalent access to the
    Corresponding Source in the same way through the same place at no
    further charge.  You need not require recipients to copy the
    Corresponding Source along with the object code.  If the place to
    copy the object code is a network server, the Corresponding Source
    may be on a different server (operated by you or a third party)
    that supports equivalent copying facilities, provided you maintain
    clear directions next to the object code saying where to find the
    Corresponding Source.  Regardless of what server hosts the
    Corresponding Source, you remain obligated to ensure that it is
    available for as long as needed to satisfy these requirements.
    e) Convey the object code using peer-to-peer transmission, provided
    you inform other peers where the object code and Corresponding
    Source of the work are being offered to the general public at no
    charge under subsection 6d.
  A separable portion of the object code, whose source code is excluded
 from the Corresponding Source as a System Library, need not be
 included in conveying the object code work.
  A "User Product" is either (1) a "consumer product", which means any
 tangible personal property which is normally used for personal, family,
 or household purposes, or (2) anything designed or sold for incorporation
 into a dwelling.  In determining whether a product is a consumer product,
 doubtful cases shall be resolved in favor of coverage.  For a particular
 product received by a particular user, "normally used" refers to a
 typical or common use of that class of product, regardless of the status
 of the particular user or of the way in which the particular user
 actually uses, or expects or is expected to use, the product.  A product
 is a consumer product regardless of whether the product has substantial
 commercial, industrial or non-consumer uses, unless such uses represent
 the only significant mode of use of the product.
  "Installation Information" for a User Product means any methods,
 procedures, authorization keys, or other information required to install
 and execute modified versions of a covered work in that User Product from
 a modified version of its Corresponding Source.  The information must
 suffice to ensure that the continued functioning of the modified object
 code is in no case prevented or interfered with solely because
 modification has been made.
  If you convey an object code work under this section in, or with, or
 specifically for use in, a User Product, and the conveying occurs as
 part of a transaction in which the right of possession and use of the
 User Product is transferred to the recipient in perpetuity or for a
 fixed term (regardless of how the transaction is characterized), the
 Corresponding Source conveyed under this section must be accompanied
 by the Installation Information.  But this requirement does not apply
 if neither you nor any third party retains the ability to install
 modified object code on the User Product (for example, the work has
 been installed in ROM).
  The requirement to provide Installation Information does not include a
 requirement to continue to provide support service, warranty, or updates
 for a work that has been modified or installed by the recipient, or for
 the User Product in which it has been modified or installed.  Access to a
 network may be denied when the modification itself materially and
 adversely affects the operation of the network or violates the rules and
 protocols for communication across the network.
  Corresponding Source conveyed, and Installation Information provided,
 in accord with this section must be in a format that is publicly
 documented (and with an implementation available to the public in
 source code form), and must require no special password or key for
 unpacking, reading or copying.
  7. Additional Terms.
  "Additional permissions" are terms that supplement the terms of this
 License by making exceptions from one or more of its conditions.
 Additional permissions that are applicable to the entire Program shall
 be treated as though they were included in this License, to the extent
 that they are valid under applicable law.  If additional permissions
 apply only to part of the Program, that part may be used separately
 under those permissions, but the entire Program remains governed by
 this License without regard to the additional permissions.
  When you convey a copy of a covered work, you may at your option
 remove any additional permissions from that copy, or from any part of
 it.  (Additional permissions may be written to require their own
 removal in certain cases when you modify the work.)  You may place
 additional permissions on material, added by you to a covered work,
 for which you have or can give appropriate copyright permission.
  Notwithstanding any other provision of this License, for material you
 add to a covered work, you may (if authorized by the copyright holders of
 that material) supplement the terms of this License with terms:
    a) Disclaiming warranty or limiting liability differently from the
    terms of sections 15 and 16 of this License; or
    b) Requiring preservation of specified reasonable legal notices or
    author attributions in that material or in the Appropriate Legal
    Notices displayed by works containing it; or
    c) Prohibiting misrepresentation of the origin of that material, or
    requiring that modified versions of such material be marked in
    reasonable ways as different from the original version; or
    d) Limiting the use for publicity purposes of names of licensors or
    authors of the material; or
    e) Declining to grant rights under trademark law for use of some
    trade names, trademarks, or service marks; or
    f) Requiring indemnification of licensors and authors of that
    material by anyone who conveys the material (or modified versions of
    it) with contractual assumptions of liability to the recipient, for
    any liability that these contractual assumptions directly impose on
    those licensors and authors.
  All other non-permissive additional terms are considered "further
 restrictions" within the meaning of section 10.  If the Program as you
 received it, or any part of it, contains a notice stating that it is
 governed by this License along with a term that is a further
 restriction, you may remove that term.  If a license document contains
 a further restriction but permits relicensing or conveying under this
 License, you may add to a covered work material governed by the terms
 of that license document, provided that the further restriction does
 not survive such relicensing or conveying.
  If you add terms to a covered work in accord with this section, you
 must place, in the relevant source files, a statement of the
 additional terms that apply to those files, or a notice indicating
 where to find the applicable terms.
  Additional terms, permissive or non-permissive, may be stated in the
 form of a separately written license, or stated as exceptions;
 the above requirements apply either way.
  8. Termination.
  You may not propagate or modify a covered work except as expressly
 provided under this License.  Any attempt otherwise to propagate or
 modify it is void, and will automatically terminate your rights under
 this License (including any patent licenses granted under the third
 paragraph of section 11).
  However, if you cease all violation of this License, then your
 license from a particular copyright holder is reinstated (a)
 provisionally, unless and until the copyright holder explicitly and
 finally terminates your license, and (b) permanently, if the copyright
 holder fails to notify you of the violation by some reasonable means
 prior to 60 days after the cessation.
  Moreover, your license from a particular copyright holder is
 reinstated permanently if the copyright holder notifies you of the
 violation by some reasonable means, this is the first time you have
 received notice of violation of this License (for any work) from that
 copyright holder, and you cure the violation prior to 30 days after
 your receipt of the notice.
  Termination of your rights under this section does not terminate the
 licenses of parties who have received copies or rights from you under
 this License.  If your rights have been terminated and not permanently
 reinstated, you do not qualify to receive new licenses for the same
 material under section 10.
  9. Acceptance Not Required for Having Copies.
  You are not required to accept this License in order to receive or
 run a copy of the Program.  Ancillary propagation of a covered work
 occurring solely as a consequence of using peer-to-peer transmission
 to receive a copy likewise does not require acceptance.  However,
 nothing other than this License grants you permission to propagate or
 modify any covered work.  These actions infringe copyright if you do
 not accept this License.  Therefore, by modifying or propagating a
 covered work, you indicate your acceptance of this License to do so.
  10. Automatic Licensing of Downstream Recipients.
  Each time you convey a covered work, the recipient automatically
 receives a license from the original licensors, to run, modify and
 propagate that work, subject to this License.  You are not responsible
 for enforcing compliance by third parties with this License.
  An "entity transaction" is a transaction transferring control of an
 organization, or substantially all assets of one, or subdividing an
 organization, or merging organizations.  If propagation of a covered
 work results from an entity transaction, each party to that
 transaction who receives a copy of the work also receives whatever
 licenses to the work the party's predecessor in interest had or could
 give under the previous paragraph, plus a right to possession of the
 Corresponding Source of the work from the predecessor in interest, if
 the predecessor has it or can get it with reasonable efforts.
  You may not impose any further restrictions on the exercise of the
 rights granted or affirmed under this License.  For example, you may
 not impose a license fee, royalty, or other charge for exercise of
 rights granted under this License, and you may not initiate litigation
 (including a cross-claim or counterclaim in a lawsuit) alleging that
 any patent claim is infringed by making, using, selling, offering for
 sale, or importing the Program or any portion of it.
  11. Patents.
  A "contributor" is a copyright holder who authorizes use under this
 License of the Program or a work on which the Program is based.  The
 work thus licensed is called the contributor's "contributor version".
  A contributor's "essential patent claims" are all patent claims
 owned or controlled by the contributor, whether already acquired or
 hereafter acquired, that would be infringed by some manner, permitted
 by this License, of making, using, or selling its contributor version,
 but do not include claims that would be infringed only as a
 consequence of further modification of the contributor version.  For
 purposes of this definition, "control" includes the right to grant
 patent sublicenses in a manner consistent with the requirements of
 this License.
  Each contributor grants you a non-exclusive, worldwide, royalty-free
 patent license under the contributor's essential patent claims, to
 make, use, sell, offer for sale, import and otherwise run, modify and
 propagate the contents of its contributor version.
  In the following three paragraphs, a "patent license" is any express
 agreement or commitment, however denominated, not to enforce a patent
 (such as an express permission to practice a patent or covenant not to
 sue for patent infringement).  To "grant" such a patent license to a
 party means to make such an agreement or commitment not to enforce a
 patent against the party.
  If you convey a covered work, knowingly relying on a patent license,
 and the Corresponding Source of the work is not available for anyone
 to copy, free of charge and under the terms of this License, through a
 publicly available network server or other readily accessible means,
 then you must either (1) cause the Corresponding Source to be so
 available, or (2) arrange to deprive yourself of the benefit of the
 patent license for this particular work, or (3) arrange, in a manner
 consistent with the requirements of this License, to extend the patent
 license to downstream recipients.  "Knowingly relying" means you have
 actual knowledge that, but for the patent license, your conveying the
 covered work in a country, or your recipient's use of the covered work
 in a country, would infringe one or more identifiable patents in that
 country that you have reason to believe are valid.
  If, pursuant to or in connection with a single transaction or
 arrangement, you convey, or propagate by procuring conveyance of, a
 covered work, and grant a patent license to some of the parties
 receiving the covered work authorizing them to use, propagate, modify
 or convey a specific copy of the covered work, then the patent license
 you grant is automatically extended to all recipients of the covered
 work and works based on it.
  A patent license is "discriminatory" if it does not include within
 the scope of its coverage, prohibits the exercise of, or is
 conditioned on the non-exercise of one or more of the rights that are
 specifically granted under this License.  You may not convey a covered
 work if you are a party to an arrangement with a third party that is
 in the business of distributing software, under which you make payment
 to the third party based on the extent of your activity of conveying
 the work, and under which the third party grants, to any of the
 parties who would receive the covered work from you, a discriminatory
 patent license (a) in connection with copies of the covered work
 conveyed by you (or copies made from those copies), or (b) primarily
 for and in connection with specific products or compilations that
 contain the covered work, unless you entered into that arrangement,
 or that patent license was granted, prior to 28 March 2007.
  Nothing in this License shall be construed as excluding or limiting
 any implied license or other defenses to infringement that may
 otherwise be available to you under applicable patent law.
  12. No Surrender of Others' Freedom.
  If conditions are imposed on you (whether by court order, agreement or
 otherwise) that contradict the conditions of this License, they do not
 excuse you from the conditions of this License.  If you cannot convey a
 covered work so as to satisfy simultaneously your obligations under this
 License and any other pertinent obligations, then as a consequence you may
 not convey it at all.  For example, if you agree to terms that obligate you
 to collect a royalty for further conveying from those to whom you convey
 the Program, the only way you could satisfy both those terms and this
 License would be to refrain entirely from conveying the Program.
  13. Use with the GNU Affero General Public License.
  Notwithstanding any other provision of this License, you have
 permission to link or combine any covered work with a work licensed
 under version 3 of the GNU Affero General Public License into a single
 combined work, and to convey the resulting work.  The terms of this
 License will continue to apply to the part which is the covered work,
 but the special requirements of the GNU Affero General Public License,
 section 13, concerning interaction through a network will apply to the
 combination as such.
  14. Revised Versions of this License.
  The Free Software Foundation may publish revised and/or new versions of
 the GNU General Public License from time to time.  Such new versions will
 be similar in spirit to the present version, but may differ in detail to
 address new problems or concerns.
  Each version is given a distinguishing version number.  If the
 Program specifies that a certain numbered version of the GNU General
 Public License "or any later version" applies to it, you have the
 option of following the terms and conditions either of that numbered
 version or of any later version published by the Free Software
 Foundation.  If the Program does not specify a version number of the
 GNU General Public License, you may choose any version ever published
 by the Free Software Foundation.
  If the Program specifies that a proxy can decide which future
 versions of the GNU General Public License can be used, that proxy's
 public statement of acceptance of a version permanently authorizes you
 to choose that version for the Program.
  Later license versions may give you additional or different
 permissions.  However, no additional obligations are imposed on any
 author or copyright holder as a result of your choosing to follow a
 later version.
  15. Disclaimer of Warranty.
  THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
 APPLICABLE LAW.  EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
 HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
 OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
 THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
 PURPOSE.  THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
 IS WITH YOU.  SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
 ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
  16. Limitation of Liability.
  IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
 WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
 THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
 GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
 USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
 DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
 PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
 EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
 SUCH DAMAGES.
  17. Interpretation of Sections 15 and 16.
  If the disclaimer of warranty and limitation of liability provided
 above cannot be given local legal effect according to their terms,
 reviewing courts shall apply local law that most closely approximates
 an absolute waiver of all civil liability in connection with the
 Program, unless a warranty or assumption of liability accompanies a
 copy of the Program in return for a fee.
                     END OF TERMS AND CONDITIONS
            How to Apply These Terms to Your New Programs
  If you develop a new program, and you want it to be of the greatest
 possible use to the public, the best way to achieve this is to make it
 free software which everyone can redistribute and change under these terms.
  To do so, attach the following notices to the program.  It is safest
 to attach them to the start of each source file to most effectively
 state the exclusion of warranty; and each file should have at least
 the "copyright" line and a pointer to where the full notice is found.
    <one line to give the program's name and a brief idea of what it does.>
    Copyright (C) <year>  <name of author>
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation, either version 3 of the License, or
    (at your option) any later version.
    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.
    You should have received a copy of the GNU General Public License
    along with this program.  If not, see <https://www.gnu.org/licenses/>.
 Also add information on how to contact you by electronic and paper mail.
  If the program does terminal interaction, make it output a short
 notice like this when it starts in an interactive mode:
    <program>  Copyright (C) <year>  <name of author>
    This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
    This is free software, and you are welcome to redistribute it
    under certain conditions; type `show c' for details.
 The hypothetical commands `show w' and `show c' should show the appropriate
 parts of the General Public License.  Of course, your program's commands
 might be different; for a GUI interface, you would use an "about box".
  You should also get your employer (if you work as a programmer) or school,
 if any, to sign a "copyright disclaimer" for the program, if necessary.
 For more information on this, and how to apply and follow the GNU GPL, see
 <https://www.gnu.org/licenses/>.
  The GNU General Public License does not permit incorporating your program
 into proprietary programs.  If your program is a subroutine library, you
 may consider it more useful to permit linking proprietary applications with
 the library.  If this is what you want to do, use the GNU Lesser General
 Public License instead of this License.  But first, please read
 <https://www.gnu.org/licenses/why-not-lgpl.html>.
@@ -7,6 +7,9 @@ A ComfyUI custom node for [OmniVoice](https://github.com/k2-fsa/OmniVoice) — a
 - **Voice Cloning** — clone any voice from a short reference audio clip
 - **Voice Design** — describe a voice with text (e.g. "female, low pitch, british accent")
 - **Auto Voice** — let the model pick a voice automatically
 - **Voice Presets** — built-in curated reference voices, ready to use without any audio file
 - **Voice Mixing** — blend two or three reference voices for a hybrid speaker
 - **EPUB Loader** — load chapters from an ebook directly into the pipeline
 - **Audiobook-ready** — handles arbitrarily long text with near-constant VRAM via built-in chunking
 - **Multilingual** — 600+ languages
@@ -33,15 +36,13 @@ A ComfyUI custom node for [OmniVoice](https://github.com/k2-fsa/OmniVoice) — a
 ### OmniVoice Model Loader
-Loads the OmniVoice model. Downloads automatically from HuggingFace on first run and caches locally.
+Loads the OmniVoice model. Downloads automatically from HuggingFace on first run and caches to `ComfyUI/models/omnivoice/`.
 | Input | Type | Description |
 |-------|------|-------------|
 | `device` | dropdown | `cuda:0`, `cuda:1`, or `cpu` |
 | `dtype` | dropdown | `float16`, `bfloat16`, or `float32` |
 Downloads automatically from HuggingFace on first run and caches to `ComfyUI/models/omnivoice/`.
 **Output:** `OMNIVOICE_MODEL`
 ---
@@ -56,25 +57,99 @@ Generates speech from text using a loaded model.
 | `text` | string | Text to synthesize (full pages supported) |
 | `mode` | dropdown | `voice_cloning`, `voice_design`, or `auto_voice` |
 | `ref_audio` | AUDIO | Reference audio for voice cloning (optional) |
-| `ref_text` | string | Transcription of ref audio — auto-detected if blank (optional) |
+| `ref_text` | string | Transcription of ref audio — connect a Whisper node for best results (optional) |
 | `instruct` | string | Voice description for voice design mode (optional) |
 | `speed` | float | Speed multiplier — default 1.0 |
 | `num_step` | int | Diffusion steps — default 32 (use 16 for faster generation) |
 | `seed` | int | Diffusion seed — set the same value across all Generate nodes in an audiobook pipeline to keep the voice consistent. 0 = random |
 **Output:** `AUDIO` at 24kHz — connects directly to ComfyUI's Save Audio node.
-## Example Workflow (Audiobook)
+---
 ### OmniVoice Voice Preset
 Pre-fetched reference voices. Audio is downloaded once and cached to `ComfyUI/models/omnivoice/presets/`.
 | Input | Type | Description |
 |-------|------|-------------|
 | `preset` | dropdown | Choose from built-in voices |
 **Outputs:** `ref_audio` (AUDIO), `ref_text` (STRING) — wire directly into OmniVoice Generate.
 Available presets:
 | Name | Gender | Style |
 |------|--------|-------|
 | Shadowheart | Female | Expressive |
 | American actress | Female | Theatrical |
 | Podcast host | Female | Casual |
 | Nature | Male | Warm |
 | Old Hollywood | Male | Classic |
 | Rick Sanchez | Male | Casual |
 | Stewie Griffin | Male | Precise |
 | Harvey Keitel | Male | Intense |
 | Conan O'Brien | Male | Comedy |
 ---
 ### OmniVoice Mix Voices
 Concatenates two or three reference audio clips to create a blended speaker. The model extracts a speaker embedding from the combined clip, producing a hybrid voice.
 | Input | Type | Description |
 |-------|------|-------------|
 | `audio_1` | AUDIO | First reference voice (required) |
 | `audio_2` | AUDIO | Second reference voice (required) |
 | `weight_1` | float | Duration weight for audio_1 (0.0–1.0) |
 | `weight_2` | float | Duration weight for audio_2 (0.0–1.0) |
 | `audio_3` | AUDIO | Third reference voice (optional) |
 | `weight_3` | float | Duration weight for audio_3 (optional) |
 | `text_1/2/3` | string | Transcripts for each clip — merged into ref_text output |
 **Outputs:** `ref_audio` (AUDIO), `ref_text` (STRING) — wire directly into OmniVoice Generate.
 Weight controls how much of each clip's duration ends up in the mix. Equal weights (1.0 / 1.0) is a good starting point.
 ---
 ### OmniVoice EPUB Loader
 Loads an EPUB file and outputs selected chapters as plain text, ready to pipe into OmniVoice Generate.
 | Input | Type | Description |
 |-------|------|-------------|
 | `epub_path` | string | Absolute path to the `.epub` file |
 | `chapter_start` | int | First chapter to include (1-indexed) |
 | `chapter_end` | int | Last chapter to include (inclusive, auto-clamped) |
 **Outputs:** `text` (STRING) — selected chapters joined by `---`, `chapter_list` (STRING) — numbered list of all chapter titles for reference.
 ## Default Workflow
 A ready-to-use workflow is included at `workflows/voice_cloning.json`:
 ```
-[OmniVoice Model Loader] ─────────────────────────┐
+[OmniVoice Model Loader] ──────────────────────────────────┐
-                                                    ▼
+                                                            ▼
-[Load Audio (narrator clip)] ──► [OmniVoice Generate] ──► [Save Audio]
+[OmniVoice Voice Preset] ──► ref_audio ──► [OmniVoice Generate] ──► [Save Audio]
-                                        ▲
+                         └──► ref_text ──►
                              text = "Page 1 content..."
                              mode = voice_cloning
 ```
-Repeat the Generate + Save Audio nodes for each page, reusing the same loader.
+Load it via ComfyUI → Load Workflow.
 ## Audiobook Pipeline
 For multi-chapter audiobooks, use the same seed across all Generate nodes to keep the voice consistent between paragraphs:
 ```
 [OmniVoice Model Loader] ──────────────────────────────────────────┐
                                                                    ▼
 [OmniVoice EPUB Loader] ──► chapter text ──► [OmniVoice Generate] ──► [Save Audio]
                                                      ▲
 [OmniVoice Voice Preset] ──► ref_audio / ref_text ──►
                              seed = 42 (fixed)
 ```
 ## Credits
@@ -1,4 +1,4 @@
-from .nodes import OmniVoiceModelLoader, OmniVoiceGenerate, OmniVoiceEpubLoader, OmniVoiceVoicePreset, OmniVoiceMixVoices
+from .nodes import OmniVoiceModelLoader, OmniVoiceGenerate, OmniVoiceEpubLoader, OmniVoiceVoicePreset, OmniVoiceMixVoices, OmniVoiceVoiceDesign, OmniVoiceSpeaker, OmniVoiceSpeakers
 NODE_CLASS_MAPPINGS = {
    "OmniVoiceModelLoader": OmniVoiceModelLoader,
@@ -6,6 +6,9 @@ NODE_CLASS_MAPPINGS = {
    "OmniVoiceEpubLoader": OmniVoiceEpubLoader,
    "OmniVoiceVoicePreset": OmniVoiceVoicePreset,
    "OmniVoiceMixVoices": OmniVoiceMixVoices,
    "OmniVoiceVoiceDesign": OmniVoiceVoiceDesign,
    "OmniVoiceSpeaker": OmniVoiceSpeaker,
    "OmniVoiceSpeakers": OmniVoiceSpeakers,
 }
 NODE_DISPLAY_NAME_MAPPINGS = {
@@ -14,6 +17,11 @@ NODE_DISPLAY_NAME_MAPPINGS = {
    "OmniVoiceEpubLoader": "OmniVoice EPUB Loader",
    "OmniVoiceVoicePreset": "OmniVoice Voice Preset",
    "OmniVoiceMixVoices": "OmniVoice Mix Voices",
    "OmniVoiceVoiceDesign": "OmniVoice Voice Design",
    "OmniVoiceSpeaker": "OmniVoice Speaker",
    "OmniVoiceSpeakers": "OmniVoice Speakers",
 }
-__all__ = ["NODE_CLASS_MAPPINGS", "NODE_DISPLAY_NAME_MAPPINGS"]
+WEB_DIRECTORY = "./web"
 __all__ = ["NODE_CLASS_MAPPINGS", "NODE_DISPLAY_NAME_MAPPINGS", "WEB_DIRECTORY"]
@@ -3,5 +3,7 @@ from .generator import OmniVoiceGenerate
 from .epub_loader import OmniVoiceEpubLoader
 from .voice_presets import OmniVoiceVoicePreset
 from .mix_voices import OmniVoiceMixVoices
 from .voice_design import OmniVoiceVoiceDesign
 from .multi_speaker import OmniVoiceSpeaker, OmniVoiceSpeakers
-__all__ = ["OmniVoiceModelLoader", "OmniVoiceGenerate", "OmniVoiceEpubLoader", "OmniVoiceVoicePreset", "OmniVoiceMixVoices"]
+__all__ = ["OmniVoiceModelLoader", "OmniVoiceGenerate", "OmniVoiceEpubLoader", "OmniVoiceVoicePreset", "OmniVoiceMixVoices", "OmniVoiceVoiceDesign", "OmniVoiceSpeaker", "OmniVoiceSpeakers"]
@@ -58,6 +58,12 @@ def _extract_chapters(epub_path):
                    if tag:
                        title = tag.get_text(strip=True)
                        break
            # Remove title/heading elements so they don't appear in the body text
            if soup.title:
                soup.title.decompose()
            for hn in ['h1', 'h2', 'h3']:
                for tag in soup.find_all(hn):
                    tag.decompose()
            for tag in soup.find_all(_BLOCK_TAGS):
                tag.append(soup.new_string('\n\n'))
            text = soup.get_text(separator='')
@@ -90,8 +96,8 @@ class OmniVoiceEpubLoader:
            },
        }
-    RETURN_TYPES = ("STRING", "STRING")
+    RETURN_TYPES = ("STRING", "STRING", "STRING")
-    RETURN_NAMES = ("text", "chapter_list")
+    RETURN_NAMES = ("text", "chapter_title", "chapter_list")
    FUNCTION = "load_epub"
    CATEGORY = "OmniVoice"
@@ -100,7 +106,7 @@ class OmniVoiceEpubLoader:
        n = len(chapters)
        if n == 0:
-            return ("", "")
+            return ("", "", "")
        start = max(1, min(chapter_start, n))
        end   = max(start, min(chapter_end, n))
@@ -111,8 +117,12 @@ class OmniVoiceEpubLoader:
            for i, ch in enumerate(chapters, 1)
        )
        # chapter_title: title of the first selected chapter (useful for file naming)
        first = chapters[start - 1]
        chapter_title = first["title"] if first["title"] else f"Chapter {start}"
        # text: selected range joined by delimiter
        selected = chapters[start - 1 : end]
        text = "\n\n---\n\n".join(ch["text"] for ch in selected)
-        return (text, chapter_list)
+        return (text, chapter_title, chapter_list)
@@ -1,8 +1,26 @@
 import re
 import tempfile
 import os
 import torch
 import soundfile as sf
 _TAG_RE = re.compile(r'^\[([^\]]+)\]:?\s*(.*)', re.DOTALL)
 def _write_tmp_wav(ref_audio):
    """Write a ComfyUI AUDIO dict to a temp WAV file. Returns the path (caller must delete)."""
    tmp = tempfile.NamedTemporaryFile(suffix=".wav", delete=False)
    tmp_path = tmp.name
    tmp.close()
    waveform = ref_audio["waveform"].squeeze(0).cpu()  # (channels, samples)
    audio_np = waveform.numpy()
    sf.write(
        tmp_path,
        audio_np[0] if audio_np.shape[0] == 1 else audio_np.T,
        int(ref_audio["sample_rate"]),
    )
    return tmp_path
 class OmniVoiceGenerate:
    @classmethod
@@ -49,12 +67,21 @@ class OmniVoiceGenerate:
                        "tooltip": (
                            "voice_cloning  – clone the voice from ref_audio (requires ref_audio)\n"
                            "voice_design   – describe a voice with the instruct field (requires instruct)\n"
-                            "auto_voice     – model picks a voice automatically"
+                            "auto_voice     – model picks a voice automatically\n"
                            "\n"
                            "Ignored when a Speakers roster is connected."
                        ),
                    },
                ),
            },
            "optional": {
                "speakers": ("OMNIVOICE_SPEAKERS", {
                    "tooltip": (
                        "Connect an OmniVoice Speakers node to enable multi-speaker generation.\n"
                        "When connected, ref_audio / instruct / mode are ignored and each paragraph\n"
                        "is routed to its assigned speaker automatically."
                    ),
                }),
                "ref_audio": ("AUDIO", {
                    "tooltip": "Reference audio clip to clone the voice from. Used in voice_cloning mode.",
                }),
@@ -65,23 +92,32 @@ class OmniVoiceGenerate:
                "instruct": ("STRING", {
                    "default": "",
                    "tooltip": (
-                        "Voice description for voice_design mode. Combine attributes freely.\n"
+                        "Voice style description. Required for voice_design mode; optional in voice_cloning\n"
                        "mode to attempt accent/style transfer on top of the cloned voice.\n"
                        "Connect the OmniVoice Voice Design node for structured input.\n"
                        "\n"
                        "GENDER:   male, female\n"
                        "AGE:      child, teenager, young adult, middle-aged, elderly\n"
-                        "PITCH:    very low, low, moderate, high, very high\n"
+                        "PITCH:    very low pitch, low pitch, moderate pitch, high pitch, very high pitch, whisper\n"
                        "STYLE:    whisper\n"
                        "\n"
-                        "ENGLISH ACCENTS (text must be English):\n"
+                        "ACCENTS (only these are supported by the model):\n"
-                        "  american, british, australian, canadian,\n"
+                        "  american accent, australian accent, british accent, canadian accent,\n"
-                        "  indian, chinese, korean, japanese, portuguese, russian\n"
+                        "  chinese accent, indian accent, japanese accent, korean accent,\n"
                        "  portuguese accent, russian accent\n"
                        "\n"
                        "EXAMPLE:  female, high pitch, british accent"
                    ),
                }),
                "guidance_scale": ("FLOAT", {
                    "default": 2.0, "min": 0.0, "max": 20.0, "step": 0.1,
                    "tooltip": (
                        "Classifier-free guidance scale. Higher = more faithful to the reference/instruct, "
                        "but can over-saturate. 2.0 is a good default."
                    ),
                }),
                "speed": ("FLOAT", {
-                    "default": 1.0, "min": 0.1, "max": 3.0, "step": 0.1,
+                    "default": 1.0, "min": 0.3, "max": 3.0, "step": 0.1,
-                    "tooltip": "Playback speed multiplier. 1.0 = normal, >1.0 = faster, <1.0 = slower.",
+                    "tooltip": "Playback speed multiplier. 1.0 = normal, >1.0 = faster, <1.0 = slower. Below 0.3 produces noise and extreme VRAM usage.",
                }),
                "num_step": ("INT", {
                    "default": 32, "min": 1, "max": 100,
@@ -104,10 +140,17 @@ class OmniVoiceGenerate:
    FUNCTION = "generate"
    CATEGORY = "OmniVoice"
-    def generate(self, model, text, mode, ref_audio=None, ref_text="", instruct="", speed=1.0, num_step=32, seed=0):
+    def generate(self, model, text, mode, speakers=None, ref_audio=None, ref_text="",
                 instruct="", guidance_scale=2.0, speed=1.0, num_step=32, seed=0):
        if seed != 0:
            torch.manual_seed(seed)
-        kwargs = {"text": text, "speed": speed, "num_step": num_step}
+
        if speakers is not None:
            return self._generate_multi_speaker(
                model, text, speakers, guidance_scale, speed, num_step
            )
        kwargs = {"text": text, "speed": speed, "num_step": num_step, "guidance_scale": guidance_scale}
        if mode == "voice_cloning" and ref_audio is None:
            raise ValueError("voice_cloning mode requires ref_audio to be connected")
@@ -115,17 +158,13 @@ class OmniVoiceGenerate:
            raise ValueError("voice_design mode requires an instruct string (e.g. 'female, low pitch')")
        if mode == "voice_cloning":
-            tmp = tempfile.NamedTemporaryFile(suffix=".wav", delete=False)
+            tmp_path = _write_tmp_wav(ref_audio)
            tmp_path = tmp.name
            tmp.close()
            try:
                ref_waveform = ref_audio["waveform"].squeeze(0).cpu()  # (channels, samples)
                audio_np = ref_waveform.numpy()
                # soundfile expects (samples,) for mono or (samples, channels) for multi-channel
                sf.write(tmp_path, audio_np[0] if audio_np.shape[0] == 1 else audio_np.T, int(ref_audio["sample_rate"]))
                kwargs["ref_audio"] = tmp_path
                if ref_text:
                    kwargs["ref_text"] = ref_text
                if instruct:
                    kwargs["instruct"] = instruct
                audio_tensors = model.generate(**kwargs)
            finally:
                try:
@@ -140,9 +179,85 @@ class OmniVoiceGenerate:
        else:  # auto_voice or fallback
            audio_tensors = model.generate(**kwargs)
-        # Concatenate chunks: each tensor is (1, T) → concat along T → (1, T_total)
+        return self._tensors_to_audio(audio_tensors)
        combined = torch.cat(audio_tensors, dim=1).cpu()  # (1, T_total) on CPU
        # ComfyUI AUDIO format: (batch, channels, samples)
        waveform = combined.unsqueeze(0)  # (1, 1, T_total)
    def _generate_multi_speaker(self, model, text, speakers_data, guidance_scale, speed, num_step):
        speaker_list = speakers_data["speakers"]
        spk_mode = speakers_data["mode"]
        label_map = {s["label"].lower(): i for i, s in enumerate(speaker_list)}
        if spk_mode == "alternate_paragraphs":
            paragraphs = [p.strip() for p in text.split("\n\n") if p.strip()]
            if not paragraphs:
                raise ValueError("OmniVoice Multi-Speaker: no paragraphs found in text.")
            segments = [
                (para, speaker_list[i % len(speaker_list)])
                for i, para in enumerate(paragraphs)
            ]
        else:  # tagged_speakers
            # In tagged mode each line that starts with [Tag] begins a new segment.
            # Continuation lines (no tag) are appended to the previous segment so
            # multi-line speeches stay together. Both \n and \n\n separators work.
            raw_segments: list[list[str]] = []
            current: list[str] = []
            for line in text.splitlines():
                line = line.strip()
                if not line:
                    continue
                if _TAG_RE.match(line):
                    if current:
                        raw_segments.append(current)
                    current = [line]
                else:
                    current.append(line)
            if current:
                raw_segments.append(current)
            if not raw_segments:
                raise ValueError("OmniVoice Multi-Speaker: no tagged segments found in text.")
            segments = []
            for lines in raw_segments:
                joined = " ".join(lines)
                m = _TAG_RE.match(joined)
                if m:
                    tag = m.group(1).strip().lower()
                    body = m.group(2).strip()
                    spk = speaker_list[label_map.get(tag, 0)]
                else:
                    body = joined
                    spk = speaker_list[0]
                if body:
                    segments.append((body, spk))
        if not segments:
            raise ValueError("OmniVoice Multi-Speaker: no text segments to generate.")
        all_chunks = []
        for para_text, spk in segments:
            tmp_path = _write_tmp_wav(spk["ref_audio"])
            try:
                kwargs = {
                    "text": para_text,
                    "ref_audio": tmp_path,
                    "speed": speed,
                    "num_step": num_step,
                    "guidance_scale": guidance_scale,
                }
                if spk["ref_text"]:
                    kwargs["ref_text"] = spk["ref_text"]
                chunks = model.generate(**kwargs)
                all_chunks.extend(chunks)
            finally:
                try:
                    os.unlink(tmp_path)
                except OSError:
                    pass
        return self._tensors_to_audio(all_chunks)
    @staticmethod
    def _tensors_to_audio(tensors):
        combined = torch.cat(tensors, dim=1).cpu()  # (1, T_total)
        waveform = combined.unsqueeze(0)             # (1, 1, T_total)
        return ({"waveform": waveform, "sample_rate": 24000},)
@@ -1,10 +1,12 @@
 import os
 import torch
 _omnivoice_import_error = None
 try:
    from omnivoice import OmniVoice
-except ImportError:
+except Exception as e:
-    OmniVoice = None  # deferred; will raise at runtime if package is missing
+    OmniVoice = None
    _omnivoice_import_error = e
 try:
    import folder_paths
@@ -33,6 +35,18 @@ class OmniVoiceModelLoader:
                    ["float16", "bfloat16", "float32"],
                    {"default": "float16"},
                ),
                "compile": (
                    "BOOLEAN",
                    {
                        "default": False,
                        "tooltip": (
                            "Run torch.compile() on the model after loading. "
                            "First generation will be slow (~30-60s warmup) while the graph is compiled, "
                            "then every subsequent generation in the session will be faster. "
                            "Recommended for audiobook pipelines. Requires PyTorch 2.0+."
                        ),
                    },
                ),
            },
        }
@@ -41,11 +55,16 @@ class OmniVoiceModelLoader:
    FUNCTION = "load_model"
    CATEGORY = "OmniVoice"
-    def load_model(self, device, dtype):
+    def load_model(self, device, dtype, compile=False):
        if OmniVoice is None:
-            raise ImportError(
+            msg = (
-                "omnivoice is not installed. Run: pip install omnivoice --no-deps"
+                "omnivoice failed to import. "
                "Install it with: pip install omnivoice --no-deps\n"
                "(On Windows embedded Python: .\\python_embeded\\python.exe -m pip install omnivoice --no-deps)\n"
            )
            if _omnivoice_import_error is not None:
                msg += f"\nOriginal error: {_omnivoice_import_error}"
            raise ImportError(msg)
        model = OmniVoice.from_pretrained(
            "k2-fsa/OmniVoice",
@@ -53,4 +72,6 @@ class OmniVoiceModelLoader:
            dtype=DTYPE_MAP[dtype],
            cache_dir=CACHE_DIR,
        )
        if compile:
            model = torch.compile(model)
        return (model,)
@@ -0,0 +1,103 @@
 class OmniVoiceSpeaker:
    """Bundle a label, reference audio, and optional transcript into a speaker slot."""
    @classmethod
    def INPUT_TYPES(cls):
        return {
            "required": {
                "label": ("STRING", {
                    "default": "Narrator",
                    "tooltip": (
                        "Name used to identify this speaker.\n"
                        "In tagged_speakers mode, prefix paragraphs with [Label]:\n"
                        "  [Narrator] Once upon a time...\n"
                        "In alternate_paragraphs mode the label is informational only."
                    ),
                }),
                "ref_audio": ("AUDIO", {
                    "tooltip": "Reference audio clip for this speaker's voice.",
                }),
            },
            "optional": {
                "ref_text": ("STRING", {
                    "default": "",
                    "tooltip": "Transcript of ref_audio. Improves cloning quality.",
                }),
            },
        }
    RETURN_TYPES = ("OMNIVOICE_SPEAKER",)
    RETURN_NAMES = ("speaker",)
    FUNCTION = "build"
    CATEGORY = "OmniVoice"
    def build(self, label, ref_audio, ref_text=""):
        return ({"label": label, "ref_audio": ref_audio, "ref_text": ref_text},)
 class OmniVoiceSpeakers:
    """Collect multiple speakers into a roster for multi-speaker generation.
    The number of speaker input slots expands dynamically when num_speakers changes
    (requires the OmniVoice web extension to be loaded by ComfyUI).
    Connect one OmniVoice Speaker node per slot.
    """
    @classmethod
    def INPUT_TYPES(cls):
        # speaker_1…speaker_8 are declared here so ComfyUI validation accepts them.
        # Visibility is controlled by the JS extension (web/multi_speaker.js):
        # only the first num_speakers slots are shown as live inputs.
        optional_speakers = {
            f"speaker_{i}": ("OMNIVOICE_SPEAKER", {})
            for i in range(1, 9)
        }
        return {
            "required": {
                "num_speakers": ("INT", {
                    "default": 2, "min": 2, "max": 8, "step": 1,
                    "tooltip": (
                        "Number of active speaker slots.\n"
                        "Changing this value adds or removes speaker_N inputs on the node."
                    ),
                }),
                "mode": (
                    ["alternate_paragraphs", "tagged_speakers"],
                    {
                        "default": "alternate_paragraphs",
                        "tooltip": (
                            "alternate_paragraphs – paragraphs (separated by blank lines) rotate\n"
                            "  through speakers in order: 1 → 2 → 3 → 1 → …\n"
                            "\n"
                            "tagged_speakers – prefix each paragraph with [Label] to assign\n"
                            "  a specific speaker. Labels must match those on the Speaker nodes.\n"
                            "  Unrecognised tags fall back to speaker 1.\n"
                            "\n"
                            "  Example:\n"
                            "  [Narrator] The door creaked open.\n"
                            "\n"
                            "  [Alice] Who is there?"
                        ),
                    },
                ),
            },
            "optional": optional_speakers,
        }
    RETURN_TYPES = ("OMNIVOICE_SPEAKERS",)
    RETURN_NAMES = ("speakers",)
    FUNCTION = "build"
    CATEGORY = "OmniVoice"
    def build(self, num_speakers, mode, **kwargs):
        speakers = []
        for i in range(1, num_speakers + 1):
            spk = kwargs.get(f"speaker_{i}")
            if spk is not None:
                speakers.append(spk)
        if len(speakers) < 2:
            raise ValueError(
                f"OmniVoice Speakers: at least 2 speakers must be connected "
                f"(got {len(speakers)})."
            )
        return ({"speakers": speakers, "mode": mode},)
@@ -0,0 +1,74 @@
 class OmniVoiceVoiceDesign:
    """Compose a voice design instruct string from structured dropdowns."""
    GENDERS = ["none", "male", "female"]
    AGES = ["none", "child", "teenager", "young adult", "middle-aged", "elderly"]
    PITCHES = [
        "none",
        "very low pitch", "low pitch", "moderate pitch",
        "high pitch", "very high pitch", "whisper",
    ]
    # Exactly the accents validated by the model's _resolve_instruct() for English
    ACCENTS = [
        "none",
        "american accent", "australian accent", "british accent",
        "canadian accent", "chinese accent", "indian accent",
        "japanese accent", "korean accent", "portuguese accent",
        "russian accent",
    ]
    # Chinese dialect items validated by the model's _resolve_instruct()
    ZH_GENDERS  = ["none", "男", "女"]
    ZH_AGES     = ["none", "儿童", "少年", "青年", "中年", "老年"]
    ZH_PITCHES  = ["none", "极低音调", "低音调", "中音调", "高音调", "极高音调", "耳语"]
    ZH_DIALECTS = [
        "none",
        "东北话", "云南话", "四川话", "宁夏话", "桂林话",
        "河南话", "济南话", "甘肃话", "石家庄话", "贵州话",
        "陕西话", "青岛话",
    ]
    @classmethod
    def INPUT_TYPES(cls):
        return {
            "required": {
                "language": (
                    ["English", "Chinese"],
                    {
                        "default": "English",
                        "tooltip": "Selects the instruct vocabulary. The language output wires directly into Generate — no need to set it there too.",
                    },
                ),
                "gender": (cls.GENDERS, {"default": "female",
                    "tooltip": "Voice gender (English). Ignored when language is Chinese — use zh_gender."}),
                "age":    (cls.AGES,    {"default": "none",
                    "tooltip": "Age of the speaker (English). Ignored when language is Chinese — use zh_age."}),
                "pitch":  (cls.PITCHES, {"default": "none",
                    "tooltip": "Pitch (English). Ignored when language is Chinese — use zh_pitch."}),
                "accent": (cls.ACCENTS, {"default": "none",
                    "tooltip": "Accent (English only, 10 supported values)."}),
            },
            "optional": {
                "zh_gender":  (cls.ZH_GENDERS,  {"default": "none", "tooltip": "声线性别 (Chinese mode)"}),
                "zh_age":     (cls.ZH_AGES,      {"default": "none", "tooltip": "年龄段 (Chinese mode)"}),
                "zh_pitch":   (cls.ZH_PITCHES,   {"default": "none", "tooltip": "音调 (Chinese mode)"}),
                "zh_dialect": (cls.ZH_DIALECTS,  {"default": "none", "tooltip": "方言/口音 (Chinese mode)"}),
            },
        }
    RETURN_TYPES  = ("STRING",)
    RETURN_NAMES  = ("instruct",)
    FUNCTION      = "compose"
    CATEGORY      = "OmniVoice"
    def compose(self, language, gender, age, pitch, accent,
                zh_gender="none", zh_age="none", zh_pitch="none", zh_dialect="none"):
        if language == "Chinese":
            parts = [v for v in [zh_gender, zh_age, zh_pitch, zh_dialect] if v != "none"]
            return ("，".join(parts),)
        else:
            parts = [v for v in [gender, age, pitch, accent] if v != "none"]
            return (", ".join(parts),)
@@ -53,14 +53,45 @@ PRESETS = {
 }
-def _load_audio(url):
+_AUDIO_EXTS = {".wav", ".flac", ".mp3", ".ogg", ".m4a"}
-    """Download (once) and return (waveform_tensor, sample_rate)."""
+_BUILTIN_FILES = frozenset(os.path.basename(url.split("?")[0]) for url, _ in PRESETS.values())
 def _scan_user_presets():
    """Return a dict of user presets found in _CACHE_DIR.
    For each audio file that is not a cached built-in, look for a same-stem
    .txt file for the transcript.  Key format: "<stem> (local)".
    """
    user = {}
    if not os.path.isdir(_CACHE_DIR):
        return user
    for fname in sorted(os.listdir(_CACHE_DIR)):
        stem, ext = os.path.splitext(fname)
        if ext.lower() not in _AUDIO_EXTS or fname in _BUILTIN_FILES:
            continue
        audio_path = os.path.join(_CACHE_DIR, fname)
        txt_path = os.path.join(_CACHE_DIR, stem + ".txt")
        transcript = ""
        if os.path.exists(txt_path):
            with open(txt_path, "r", encoding="utf-8") as f:
                transcript = f.read().strip()
        user[f"{stem} (local)"] = (audio_path, transcript)
    return user
 def _load_audio(source):
    """Load audio from a URL (downloading once) or a local file path."""
    os.makedirs(_CACHE_DIR, exist_ok=True)
-    filename = os.path.basename(url.split("?")[0])
+    if source.startswith("http://") or source.startswith("https://"):
-    cache_path = os.path.join(_CACHE_DIR, filename)
+        filename = os.path.basename(source.split("?")[0])
-    if not os.path.exists(cache_path):
+        cache_path = os.path.join(_CACHE_DIR, filename)
-        urllib.request.urlretrieve(url, cache_path)
+        if not os.path.exists(cache_path):
-    audio_np, sr = sf.read(cache_path, dtype="float32")
+            urllib.request.urlretrieve(source, cache_path)
        path = cache_path
    else:
        path = source
    audio_np, sr = sf.read(path, dtype="float32")
    if audio_np.ndim == 1:
        audio_np = audio_np[np.newaxis, :]        # (1, samples)
    else:
@@ -72,15 +103,20 @@ def _load_audio(url):
 class OmniVoiceVoicePreset:
    @classmethod
    def INPUT_TYPES(cls):
        all_presets = {**PRESETS, **_scan_user_presets()}
        return {
            "required": {
                "preset": (
-                    list(PRESETS.keys()),
+                    list(all_presets.keys()),
                    {
                        "tooltip": (
                            "Pre-fetched reference voice for OmniVoice Generate.\n"
                            "Connect ref_audio → ref_audio and ref_text → ref_text.\n"
-                            "If ref_text is blank, connect a Whisper node to supply the transcript."
+                            "\n"
                            "To add your own presets, drop audio files into:\n"
                            f"  {_CACHE_DIR}\n"
                            "Add a same-name .txt file alongside for the transcript.\n"
                            "Restart ComfyUI to pick up new files."
                        ),
                    },
                ),
@@ -93,6 +129,7 @@ class OmniVoiceVoicePreset:
    CATEGORY = "OmniVoice"
    def load_preset(self, preset):
-        url, transcript = PRESETS[preset]
+        all_presets = {**PRESETS, **_scan_user_presets()}
-        waveform, sr = _load_audio(url)
+        source, transcript = all_presets[preset]
        waveform, sr = _load_audio(source)
        return ({"waveform": waveform, "sample_rate": sr}, transcript)
@@ -0,0 +1,14 @@
 [project]
 name = "comfyui-omnivoice-fel"
 description = "ComfyUI nodes for OmniVoice — multilingual zero-shot TTS with voice cloning, voice design, EPUB loading, and voice mixing."
 version = "1.0.6"
 license = { text = "GPL-3.0-only" }
 dependencies = []
 [project.urls]
 Repository = "https://github.com/ethanfel/ComfyUI-Omnivoice"
 [tool.comfy]
 PublisherId = "ethanfel"
 DisplayName = "ComfyUI-Omnivoice"
 Icon = ""
@@ -1,6 +1,9 @@
-transformers>=4.40.0
+transformers>=5.3.0
 accelerate
 pydub
 soundfile
 numpy
 pydub
 tensorboardx
 webdataset
 beautifulsoup4
 torchcodec
@@ -58,14 +58,14 @@ def test_input_types_structure():
 def test_return_types():
-    assert OmniVoiceEpubLoader.RETURN_TYPES == ("STRING", "STRING")
+    assert OmniVoiceEpubLoader.RETURN_TYPES == ("STRING", "STRING", "STRING")
-    assert OmniVoiceEpubLoader.RETURN_NAMES == ("text", "chapter_list")
+    assert OmniVoiceEpubLoader.RETURN_NAMES == ("text", "chapter_title", "chapter_list")
 def test_chapter_extraction_basic():
    epub = make_fake_epub([("Intro", "<p>Hello world</p>"), ("Chapter One", "<p>Body here</p>")])
    with patch('nodes.epub_loader.zipfile.ZipFile', side_effect=epub_opener(epub)):
-        text, chapter_list = OmniVoiceEpubLoader().load_epub('/fake.epub', 1, 2)
+        text, chapter_title, chapter_list = OmniVoiceEpubLoader().load_epub('/fake.epub', 1, 2)
    assert "Hello world" in text
    assert "Body here" in text
    assert "---" in text
@@ -75,7 +75,7 @@ def test_chapter_extraction_basic():
 def test_chapter_range_single():
    epub = make_fake_epub([("One", "<p>First</p>"), ("Two", "<p>Second</p>"), ("Three", "<p>Third</p>")])
    with patch('nodes.epub_loader.zipfile.ZipFile', side_effect=epub_opener(epub)):
-        text, _ = OmniVoiceEpubLoader().load_epub('/fake.epub', 2, 2)
+        text, _, _ = OmniVoiceEpubLoader().load_epub('/fake.epub', 2, 2)
    assert "Second" in text
    assert "First" not in text
    assert "Third" not in text
@@ -84,7 +84,7 @@ def test_chapter_range_single():
 def test_chapter_list_contains_all():
    epub = make_fake_epub([("A", ""), ("B", ""), ("C", "")])
    with patch('nodes.epub_loader.zipfile.ZipFile', side_effect=epub_opener(epub)):
-        _, chapter_list = OmniVoiceEpubLoader().load_epub('/fake.epub', 2, 2)
+        _, _, chapter_list = OmniVoiceEpubLoader().load_epub('/fake.epub', 2, 2)
    lines = chapter_list.strip().splitlines()
    assert len(lines) == 3
    assert lines[0].startswith("1.")
@@ -94,14 +94,14 @@ def test_chapter_list_contains_all():
 def test_range_clamping_high():
    epub = make_fake_epub([("A", "<p>aaa</p>"), ("B", "<p>bbb</p>")])
    with patch('nodes.epub_loader.zipfile.ZipFile', side_effect=epub_opener(epub)):
-        text, _ = OmniVoiceEpubLoader().load_epub('/fake.epub', 1, 99)
+        text, _, _ = OmniVoiceEpubLoader().load_epub('/fake.epub', 1, 99)
    assert "aaa" in text and "bbb" in text
 def test_range_clamping_end_below_start():
    epub = make_fake_epub([("A", "<p>aaa</p>"), ("B", "<p>bbb</p>")])
    with patch('nodes.epub_loader.zipfile.ZipFile', side_effect=epub_opener(epub)):
-        text, _ = OmniVoiceEpubLoader().load_epub('/fake.epub', 2, 1)
+        text, _, _ = OmniVoiceEpubLoader().load_epub('/fake.epub', 2, 1)
    assert "bbb" in text
    assert "aaa" not in text
@@ -119,14 +119,14 @@ def test_missing_title_fallback():
        z.writestr('OEBPS/ch0.xhtml', '<html><body><p>No title here</p></body></html>')
    buf.seek(0)
    with patch('nodes.epub_loader.zipfile.ZipFile', side_effect=epub_opener(buf.read())):
-        _, chapter_list = OmniVoiceEpubLoader().load_epub('/fake.epub', 1, 1)
+        _, _, chapter_list = OmniVoiceEpubLoader().load_epub('/fake.epub', 1, 1)
    assert "1. Chapter 1" in chapter_list
 def test_script_style_stripped():
    epub = make_fake_epub([("Test", '<script>alert("xss")</script><style>color:red</style><p>clean</p>')])
    with patch('nodes.epub_loader.zipfile.ZipFile', side_effect=epub_opener(epub)):
-        text, _ = OmniVoiceEpubLoader().load_epub('/fake.epub', 1, 1)
+        text, _, _ = OmniVoiceEpubLoader().load_epub('/fake.epub', 1, 1)
    assert "alert" not in text
    assert "color" not in text
    assert "clean" in text
@@ -46,7 +46,7 @@ def test_generate_auto_voice():
    assert "sample_rate" in audio
    assert audio["sample_rate"] == 24000
    mock_model.generate.assert_called_once_with(
-        text="Hello world", speed=1.0, num_step=32
+        text="Hello world", speed=1.0, num_step=32, guidance_scale=2.0
    )
@@ -64,7 +64,7 @@ def test_generate_voice_design():
    audio = result[0]
    assert audio["sample_rate"] == 24000
    mock_model.generate.assert_called_once_with(
-        text="Hello world", instruct="female, low pitch", speed=1.0, num_step=32
+        text="Hello world", instruct="female, low pitch", speed=1.0, num_step=32, guidance_scale=2.0
    )
@@ -0,0 +1,70 @@
 import { app } from "../../scripts/app.js";
 const MAX_SPEAKERS = 8;
 app.registerExtension({
    name: "OmniVoice.MultiSpeaker",
    beforeRegisterNodeDef(nodeType, nodeData) {
        if (nodeData.name !== "OmniVoiceSpeakers") return;
        /**
         * Ensure the node has exactly `count` speaker_N inputs.
         * Safe to call multiple times with the same count (idempotent).
         */
        function syncSpeakerInputs(node, count) {
            count = Math.max(2, Math.min(MAX_SPEAKERS, Math.floor(count)));
            // Add any missing slots
            for (let i = 1; i <= count; i++) {
                const name = `speaker_${i}`;
                if (!node.inputs?.find(inp => inp.name === name)) {
                    node.addInput(name, "OMNIVOICE_SPEAKER");
                }
            }
            // Remove excess slots (high → low so indices stay valid)
            for (let i = MAX_SPEAKERS; i > count; i--) {
                const name = `speaker_${i}`;
                const idx = node.inputs?.findIndex(inp => inp.name === name) ?? -1;
                if (idx === -1) continue;
                // Sever any connected link before removing the slot
                const linkId = node.inputs[idx].link;
                if (linkId != null) node.graph?.removeLink(linkId);
                node.removeInput(idx);
            }
            node.setDirtyCanvas(true, true);
        }
        /**
         * Attach the num_speakers widget callback once per node instance.
         * Guarded by a flag so configure() can call it safely on reload.
         */
        function attachCallback(node) {
            if (node._omnivoiceCbAttached) return;
            const w = node.widgets?.find(w => w.name === "num_speakers");
            if (!w) return;
            node._omnivoiceCbAttached = true;
            w.callback = (value) => syncSpeakerInputs(node, value);
        }
        // --- Fresh node creation ---
        const onNodeCreated = nodeType.prototype.onNodeCreated;
        nodeType.prototype.onNodeCreated = function () {
            onNodeCreated?.apply(this, arguments);
            attachCallback(this);
            const w = this.widgets?.find(w => w.name === "num_speakers");
            if (w) syncSpeakerInputs(this, w.value);
        };
        // --- Workflow load: called by LiteGraph after widget values are restored ---
        const onConfigure = nodeType.prototype.onConfigure;
        nodeType.prototype.onConfigure = function (data) {
            onConfigure?.apply(this, arguments);
            attachCallback(this);
            const w = this.widgets?.find(w => w.name === "num_speakers");
            if (w) syncSpeakerInputs(this, w.value);
        };
    },
 });
@@ -0,0 +1,296 @@
 {
  "id": "85925298-bd01-4df7-b8e5-ac92f9d6582e",
  "revision": 0,
  "last_node_id": 14,
  "last_link_id": 30,
  "nodes": [
    {
      "id": 6,
      "type": "PreviewAudio",
      "pos": [
        928,
        96
      ],
      "size": [
        448,
        96
      ],
      "flags": {},
      "order": 4,
      "mode": 0,
      "inputs": [
        {
          "name": "audio",
          "type": "AUDIO",
          "link": 27
        }
      ],
      "outputs": [],
      "properties": {
        "cnr_id": "comfy-core",
        "ver": "0.18.1",
        "Node name for S&R": "PreviewAudio",
        "ue_properties": {
          "widget_ue_connectable": {},
          "input_ue_unconnectable": {},
          "version": "7.8"
        }
      },
      "widgets_values": []
    },
    {
      "id": 14,
      "type": "OmniVoiceGenerate",
      "pos": [
        512,
        96
      ],
      "size": [
        384,
        448
      ],
      "flags": {},
      "order": 3,
      "mode": 0,
      "inputs": [
        {
          "name": "model",
          "type": "OMNIVOICE_MODEL",
          "link": 23
        },
        {
          "name": "ref_audio",
          "shape": 7,
          "type": "AUDIO",
          "link": 29
        },
        {
          "name": "ref_text",
          "shape": 7,
          "type": "STRING",
          "widget": {
            "name": "ref_text"
          },
          "link": 30
        }
      ],
      "outputs": [
        {
          "name": "audio",
          "type": "AUDIO",
          "links": [
            27
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-omnivoice-fel",
        "ver": "340c0aa402fdf8b56fad5eb1559ff901c7fc7cfc",
        "Node name for S&R": "OmniVoiceGenerate",
        "ue_properties": {
          "widget_ue_connectable": {},
          "input_ue_unconnectable": {},
          "version": "7.8"
        }
      },
      "widgets_values": [
        "Hello! This is a test of the OmniVoice text-to-speech system.",
        "voice_cloning",
        "",
        "",
        2,
        1,
        32,
        42,
        "fixed"
      ]
    },
    {
      "id": 2,
      "type": "OmniVoiceVoicePreset",
      "pos": [
        64,
        320
      ],
      "size": [
        320,
        128
      ],
      "flags": {},
      "order": 0,
      "mode": 0,
      "inputs": [],
      "outputs": [
        {
          "name": "ref_audio",
          "type": "AUDIO",
          "slot_index": 0,
          "links": [
            29
          ]
        },
        {
          "name": "ref_text",
          "type": "STRING",
          "slot_index": 1,
          "links": [
            30
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-omnivoice-fel",
        "ver": "9b62c9bda8469171025709e2fae3ce877ee0f059",
        "Node name for S&R": "OmniVoiceVoicePreset",
        "aux_id": "ethanfel/ComfyUI-Omnivoice",
        "ue_properties": {
          "widget_ue_connectable": {},
          "version": "7.8",
          "input_ue_unconnectable": {}
        }
      },
      "widgets_values": [
        "Shadowheart – female, expressive (Chatterbox)"
      ]
    },
    {
      "id": 1,
      "type": "OmniVoiceModelLoader",
      "pos": [
        64,
        96
      ],
      "size": [
        320,
        160
      ],
      "flags": {},
      "order": 1,
      "mode": 0,
      "inputs": [],
      "outputs": [
        {
          "name": "model",
          "type": "OMNIVOICE_MODEL",
          "slot_index": 0,
          "links": [
            23
          ]
        }
      ],
      "properties": {
        "cnr_id": "comfyui-omnivoice-fel",
        "ver": "9b62c9bda8469171025709e2fae3ce877ee0f059",
        "Node name for S&R": "OmniVoiceModelLoader",
        "aux_id": "ethanfel/ComfyUI-Omnivoice",
        "ue_properties": {
          "widget_ue_connectable": {},
          "version": "7.8",
          "input_ue_unconnectable": {}
        }
      },
      "widgets_values": [
        "cuda:0",
        "float32",
        false
      ]
    },
    {
      "id": 13,
      "type": "OmniVoiceVoiceDesign",
      "pos": [
        64,
        512
      ],
      "size": [
        288,
        288
      ],
      "flags": {},
      "order": 2,
      "mode": 0,
      "inputs": [],
      "outputs": [
        {
          "name": "instruct",
          "type": "STRING",
          "links": []
        }
      ],
      "properties": {
        "cnr_id": "comfyui-omnivoice-fel",
        "ver": "340c0aa402fdf8b56fad5eb1559ff901c7fc7cfc",
        "Node name for S&R": "OmniVoiceVoiceDesign",
        "ue_properties": {
          "widget_ue_connectable": {},
          "input_ue_unconnectable": {},
          "version": "7.8"
        }
      },
      "widgets_values": [
        "English",
        "none",
        "none",
        "none",
        "american accent",
        "none",
        "none",
        "none",
        "none"
      ]
    }
  ],
  "links": [
    [
      23,
      1,
      0,
      14,
      0,
      "OMNIVOICE_MODEL"
    ],
    [
      27,
      14,
      0,
      6,
      0,
      "AUDIO"
    ],
    [
      29,
      2,
      0,
      14,
      1,
      "AUDIO"
    ],
    [
      30,
      2,
      1,
      14,
      2,
      "STRING"
    ]
  ],
  "groups": [],
  "config": {},
  "extra": {
    "ds": {
      "scale": 1.201632379715383,
      "offset": [
        1415.539606763376,
        123.61214765199416
      ]
    },
    "ue_links": [],
    "links_added_by_ue": [],
    "frontendVersion": "1.42.8",
    "VHS_latentpreview": true,
    "VHS_latentpreviewrate": 0,
    "VHS_MetadataImage": true,
    "VHS_KeepIntermediate": true
  },
  "version": 0.4
 }
@@ -1,85 +0,0 @@
 {
  "last_node_id": 4,
  "last_link_id": 4,
  "nodes": [
    {
      "id": 1,
      "type": "OmniVoiceModelLoader",
      "pos": [40, 80],
      "size": {"0": 300, "1": 100},
      "flags": {},
      "order": 0,
      "mode": 0,
      "outputs": [
        {"name": "model", "type": "OMNIVOICE_MODEL", "links": [1], "shape": 3, "slot_index": 0}
      ],
      "properties": {"Node name for S&R": "OmniVoiceModelLoader"},
      "widgets_values": ["cuda:0", "float16"]
    },
    {
      "id": 2,
      "type": "OmniVoiceVoicePreset",
      "pos": [40, 240],
      "size": {"0": 300, "1": 80},
      "flags": {},
      "order": 1,
      "mode": 0,
      "outputs": [
        {"name": "ref_audio", "type": "AUDIO",  "links": [2], "shape": 3, "slot_index": 0},
        {"name": "ref_text",  "type": "STRING", "links": [3], "shape": 3, "slot_index": 1}
      ],
      "properties": {"Node name for S&R": "OmniVoiceVoicePreset"},
      "widgets_values": ["Shadowheart – female, expressive (Chatterbox)"]
    },
    {
      "id": 3,
      "type": "OmniVoiceGenerate",
      "pos": [420, 80],
      "size": {"0": 380, "1": 340},
      "flags": {},
      "order": 2,
      "mode": 0,
      "inputs": [
        {"name": "model",     "type": "OMNIVOICE_MODEL", "link": 1, "slot_index": 0},
        {"name": "ref_audio", "type": "AUDIO",           "link": 2, "slot_index": 1},
        {"name": "ref_text",  "type": "STRING",          "link": 3, "slot_index": 2}
      ],
      "outputs": [
        {"name": "audio", "type": "AUDIO", "links": [4], "shape": 3, "slot_index": 0}
      ],
      "properties": {"Node name for S&R": "OmniVoiceGenerate"},
      "widgets_values": [
        "Hello! This is a test of the OmniVoice text-to-speech system.",
        "voice_cloning",
        "",
        1.0,
        32,
        42
      ]
    },
    {
      "id": 4,
      "type": "SaveAudio",
      "pos": [880, 80],
      "size": {"0": 300, "1": 100},
      "flags": {},
      "order": 3,
      "mode": 0,
      "inputs": [
        {"name": "audio", "type": "AUDIO", "link": 4, "slot_index": 0}
      ],
      "properties": {"Node name for S&R": "SaveAudio"},
      "widgets_values": ["omnivoice"]
    }
  ],
  "links": [
    [1, 1, 0, 3, 0, "OMNIVOICE_MODEL"],
    [2, 2, 0, 3, 1, "AUDIO"],
    [3, 2, 1, 3, 2, "STRING"],
    [4, 3, 0, 4, 0, "AUDIO"]
  ],
  "groups": [],
  "config": {},
  "extra": {"ds": {"scale": 0.9, "offset": [0, 0]}},
  "version": 0.4
 }
Author	SHA1	Message	Date
Ethanfel	b4c1cb2955	chore: bump version to 1.0.6 Publish to ComfyUI Registry / publish (push) Has been cancelled Details Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-04-10 22:17:21 +02:00
Ethanfel	2e3c357e5a	fix: raise speed minimum from 0.1 to 0.3 to prevent VRAM explosion Speed values below 0.3 produce noise, and 0.1 generates 10x the normal audio length which can consume 20+ GB VRAM and freeze the system. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-04-10 22:12:04 +02:00
Ethanfel	aa986fd534	fix: set transformers>=5.3.0 as required by omnivoice Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-04-10 12:14:13 +02:00
Ethanfel	3fee610050	fix: bump transformers minimum to >=4.57.0 Confirmed working version. Previous >=4.40.0 was too permissive and older versions may lack APIs that omnivoice depends on. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-04-10 12:13:19 +02:00
Ethanfel	d4638aa785	chore: bump version to 1.0.5 Publish to ComfyUI Registry / publish (push) Has been cancelled Details Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-04-10 12:08:44 +02:00
Ethanfel	e9c947b613	fix: strip title and heading tags from EPUB text output Publish to ComfyUI Registry / publish (push) Has been cancelled Details The chapter title was appearing multiple times in the text (from <title>, <h1>, and body). Now <title> and <h1>/<h2>/<h3> are removed from the body text since the title is already available via the chapter_title output. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-04-10 00:07:00 +02:00
Ethanfel	197bcc554e	chore: bump version to 1.0.4 Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-04-09 23:32:03 +02:00
Ethanfel	9f2683cd54	fix: add torchcodec to requirements for torchaudio backend Newer torchaudio versions default to the TorchCodec backend for loading audio files. Without it installed, omnivoice fails with ImportError when loading reference audio. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-04-09 23:31:40 +02:00
Ethanfel	f8657aca80	fix: update tests for chapter_title output and guidance_scale param - epub loader tests: unpack 3 return values (added chapter_title) - generator tests: expect guidance_scale in model.generate() calls Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-09 19:06:38 +02:00
Ethanfel	0f2bcc4c0e	chore: bump version to 1.0.3 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-09 17:41:02 +02:00
Ethanfel	4eabac4c7e	feat: add chapter_title output to EPUB Loader for file naming Returns the title of the first selected chapter as a STRING so it can be wired directly into a Save Audio node's filename field. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-09 17:40:38 +02:00
Ethanfel	dd6d5061e4	chore: bump version to 1.0.2 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-08 12:26:26 +02:00
Ethanfel	147030e2af	fix: surface real import error when omnivoice fails to load The broad except ImportError was swallowing the actual failure reason (e.g. a missing transitive dep after --no-deps install). Now captures and re-raises the original exception in the error message so users can diagnose what is actually missing. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-08 12:23:59 +02:00
Ethanfel	aedbe2e7d9	fix: declare speaker_1..8 in INPUT_TYPES so ComfyUI validation accepts them Dynamic JS inputs that are not listed in INPUT_TYPES may be rejected by ComfyUI's prompt validator and not passed to the Python function. Declaring all 8 slots as optional fixes this while JS still controls which slots are visible on the node. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-06 09:41:56 +02:00
Ethanfel	26295e4db7	feat: auto-discover user presets from the presets folder Drop any audio file (wav/flac/mp3/ogg/m4a) into the presets cache dir and it will appear as "<name> (local)" in the Voice Preset dropdown on next ComfyUI restart. Add a same-stem .txt file for the transcript. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-06 09:37:17 +02:00
Ethanfel	d5f2632c48	feat: support [tag]: syntax in tagged_speakers mode Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-06 09:21:27 +02:00
Ethanfel	33b3d62d02	fix: tagged_speakers splits on single newlines, not just double newlines Each line starting with [Tag] now begins a new segment so users don't need blank lines between tagged speeches. Continuation lines (no tag) are joined to the previous tagged segment for multi-line speeches. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-06 09:16:54 +02:00
Ethanfel	95cf706b19	feat: add multi-speaker generation with JS-powered dynamic slots - Add OmniVoiceSpeaker node (label + ref_audio + ref_text → OMNIVOICE_SPEAKER) - Add OmniVoiceSpeakers node (roster with dynamic speaker_N inputs driven by num_speakers INT widget; slots expand/collapse via ComfyUI JS extension) - Add web/multi_speaker.js: ComfyUI extension that hooks onNodeCreated and onConfigure to sync speaker_N inputs in real time (max 8 speakers) - Extend OmniVoiceGenerate with optional speakers (OMNIVOICE_SPEAKERS) input; when connected it routes each paragraph to the assigned speaker and concatenates the results — supports alternate_paragraphs and tagged_speakers modes - Remove OmniVoiceMultiSpeakerGenerate (generation now lives in the existing Generate node) - Refactor generator.py: extract _write_tmp_wav helper, add _tensors_to_audio Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-06 09:08:23 +02:00
Ethanfel	3cbc04d12d	wf: update default workflow from ComfyUI export Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:50:52 +02:00
Ethanfel	340c0aa402	simplify: remove language param entirely — model detects from instruct string Chinese characters vs English words are self-identifying to the model. No need for a separate language signal on either node. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:44:27 +02:00
Ethanfel	2b13e55dc5	fix: pipe language out of Voice Design into Generate Voice Design now outputs (instruct, language) — wire language directly into Generate to avoid setting it in two places. Generate's language input is now a STRING (accepts the connection or manual 'auto'). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:40:35 +02:00
Ethanfel	86ec8cf3fb	bump: version 1.0.1 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:33:55 +02:00
Ethanfel	ae2255d9e4	fix: add missing omnivoice runtime deps for fresh installs pydub, tensorboardx, webdataset are omnivoice dependencies that won't be present on a clean ComfyUI install since we use --no-deps. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:33:19 +02:00
Ethanfel	d5a0ebeb9a	revert: restore working requirements.txt Pinning transformers==5.3.0 risks conflicting with existing ComfyUI venv. Back to permissive >=4.40.0 which worked in practice. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:29:04 +02:00
Ethanfel	0d43e5374f	fix: sync requirements.txt with omnivoice's actual --no-deps dependencies Pins transformers==5.3.0 (omnivoice requires exact version), restores pydub (omnivoice dep), adds tensorboardx and webdataset. Drops gradio (demo-only, not needed for inference). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:28:35 +02:00
Ethanfel	2b4b221e88	chore: remove unused pydub from requirements Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:27:34 +02:00
Ethanfel	772f6654d4	feat: add language selector for voice_design + Chinese instruct support - Generate: language dropdown (auto/English/Chinese), passed only in voice_design and auto_voice modes where it selects the instruct vocab - VoiceDesign: Chinese mode with dialect/age/pitch/gender dropdowns using the model's validated Chinese instruct vocabulary (全角逗号) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:22:25 +02:00
Ethanfel	e26bac3684	remove: language parameter from Generate (model auto-detects from text) Language is inferred from the text content — the parameter had no effect. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:21:27 +02:00
Ethanfel	194e0b0e09	fix: trim accent list to model-validated values only The model's _resolve_instruct() validates against a fixed vocabulary. Only 10 accents are supported — removed all unsupported additions. Updated tooltip to reflect actual constraints. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:17:59 +02:00
Ethanfel	d4bf7c825e	feat: pass instruct in voice_cloning mode for accent/style influence If instruct is set alongside ref_audio, it is now forwarded to model.generate() — allowing accent/style transfer on top of the cloned voice identity. Model may or may not honour both simultaneously. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:07:18 +02:00
Ethanfel	d2cb5c4249	feat: expand language and accent lists to full coverage Language: ~170 world languages with type-to-filter dropdown Accent: 50+ regional varieties grouped by area Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:04:12 +02:00
Ethanfel	c1558efad9	feat: add Voice Design node + language and guidance_scale to Generate OmniVoiceVoiceDesign: structured dropdowns for gender/age/pitch/accent that compose into an instruct string — wire to Generate's instruct input. OmniVoiceGenerate: new optional language dropdown (auto + 11 languages) and guidance_scale (CFG, default 2.0) parameters. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 20:02:06 +02:00
Ethanfel	97ed0f209f	fix: rename node id to comfyui-omnivoice-fel (comfyui-omnivoice is taken) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:59:17 +02:00
Ethanfel	f7d624799c	fix: revert PublisherId to lowercase ethanfel (matches registry) Publish to ComfyUI Registry / publish (push) Has been cancelled Details Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:46:55 +02:00
Ethanfel	d5000dee11	fix: correct PublisherId casing to Ethanfel Publish to ComfyUI Registry / publish (push) Has been cancelled Details Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:40:52 +02:00
Ethanfel	bb1d83578c	ci: trigger publish on pyproject.toml push to master (not tags) Matches the pattern used by the working LoRA Optimizer publish workflow. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:37:34 +02:00
Ethanfel	138df3abc4	fix: align pyproject.toml format with working registry publish Publish to ComfyUI Registry / publish (push) Has been cancelled Details - license: use { text = "GPL-3.0-only" } (matches comfy-cli expectation) - add dependencies = [] Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:35:31 +02:00
Ethanfel	db884341e8	fix: move repository URL to [project.urls] for comfy-cli Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:34:30 +02:00
Ethanfel	dfad38a5c3	fix: correct pyproject.toml for registry publish Publish to ComfyUI Registry / publish (push) Has been cancelled Details - license: use SPDX identifier GPL-3.0-only instead of file reference - add Repository URL Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:32:52 +02:00
Ethanfel	f4697ae88f	wf: update default workflow from ComfyUI export Publish to ComfyUI Registry / publish (push) Has been cancelled Details Refreshed node IDs, positions and sizes from live session. Replaced SaveAudio with PreviewAudio, added ref_text widget entry, updated aux_id/ver properties. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:27:45 +02:00
Ethanfel	616c2a3e61	license: add GNU GPL v3.0 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:27:22 +02:00
Ethanfel	9b62c9bda8	fix: add seed control_after_generate value to default workflow ComfyUI appends a hidden "fixed"/"randomize" value after every INT named "seed". Without it the widget values were misaligned. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:21:13 +02:00
Ethanfel	46591553f9	feat: add torch.compile option to Model Loader Compiles the model graph on first generation (~30-60s warmup) then speeds up all subsequent generations in the session. Recommended for audiobook pipelines. Default off. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:16:39 +02:00
Ethanfel	2e1cba61fc	fix: adjust node sizes in default workflow for seed widget Generate node height 340→400 to fit all 6 widgets, Voice Preset height 80→100, SaveAudio position adjusted. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:12:31 +02:00
Ethanfel	c22cc7c296	rename: voice_cloning.json → omnivoice_voice_cloning.json Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:10:51 +02:00
Ethanfel	f7756e6240	ci: add pyproject.toml for ComfyUI registry Sets publisher ID, display name, and package metadata required for Comfy-Org/publish-node-action. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:10:20 +02:00
Ethanfel	8e2367f8f0	ci: add ComfyUI registry publish workflow Triggers on version tags (v*) using Comfy-Org/publish-node-action. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:09:40 +02:00
Ethanfel	14542a6b00	docs: update README with all nodes (presets, mix voices, EPUB loader) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-05 19:07:15 +02:00