Ethanfel/ComfyUI-UniverSR
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: auto input_sr — detect bandwidth and pick the best value

Browse Source 
New "auto" option (now the default) on the Sampler's input_sr. detect_input_sr
finds the spectral cutoff cliff (steepest drop) and its dB confidence: effective
cutoff = that cliff if confident, else sr/2 — one rule that covers band-limited
(→ matched input_sr), full-band (→ 24000), and genuine low-rate files
(→ their rate). Rounds DOWN to the nearest supported Nyquist to avoid feeding
the model an empty band. Logs its decision. Falls back to 24000 when unsure.

Tests cover sharp 4/6/8/12 kHz cutoffs, full-band, genuine-8kHz, silence, stereo.
Verified end-to-end on the real model (8 kHz clip -> auto picks 16000).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
Ethanfel
2026-06-17 12:46:02 +02:00
parent 8d4cd71723
commit e5110b88e1
4 changed files with 165 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+8 -1
View File
@@ -142,7 +142,7 @@ Runs the super-resolution. Outputs: **`AUDIO`** (48 kHz) and **`IMAGE`** (spectr
|---|---|---|---|---|
| `audio` | AUDIO | — | — | Input audio (any sample rate / mono or stereo). |
| `model` | UNIVERSR_MODEL | — | — | From the Model Loader. |
| `input_sr` | choice | `8000` | 8000 / 12000 / 16000 / 24000 | **Effective input bandwidth (Hz).** Content is treated as valid up to `input_sr/2` and **regenerated above it**. See below. |
| `input_sr` | choice | `auto` | auto / 8000 / 12000 / 16000 / 24000 | **Effective input bandwidth (Hz).** Content is valid up to `input_sr/2` and **regenerated above it**. `auto` detects the cutoff for you (see below). |
| `ode_method` | choice | `midpoint` | euler / midpoint / rk4 | ODE solver. `euler` fastest → `midpoint` balanced → `rk4` best. |
| `ode_steps` | int | `4` | 1–64 | Flow-matching integration steps. `4` is fast & validated; `4–10` is a good range. |
| `guidance_scale` | float | `1.5` | 0–6 | Classifier-free guidance. Higher = denser highs but less faithful. `0` disables CFG. |
@@ -210,6 +210,13 @@ audio **and** the `video` reference into the combiner. Ready-made graph:
| `16000` | 8 kHz | 8 – 24 kHz |
| `24000` | 12 kHz | 12 – 24 kHz |
**`auto` (default)** analyses the input's spectrum, finds the **cutoff cliff**, and picks the largest
supported bandwidth at or below it (rounding *down*, to avoid feeding the model an empty band). It
prints its decision, e.g. `auto: cutoff 8.0 kHz (drop 53 dB) -> input_sr=16000`. When there's **no clear
cutoff** (full-band or gently rolled-off audio) it falls back to `24000` (least aggressive). Auto is
most reliable on genuinely band-limited material (codecs, downsamples, telephone); for fine control or
deliberate over-brightening, pick a value manually.
Two ways to use it:
1. **Genuine low-rate audio (classic super-resolution).** You have an 8 kHz (or 16/24 kHz) recording