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

fix: lower RMS normalization target from -23/-20 to -27 dBFS

Browse Source 
Training clips at -23 LUFS measure -25 to -31 dBFS RMS (avg ~-27).
Normalizing output to -23 dBFS was 4-8 dB too loud, causing saturation
on clips with high crest factor and peaks near 0 dBFS.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Ethanfel
2026-04-08 17:19:20 +02:00
parent 678c050f11
commit 9a47508d2d
2 changed files with 4 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
nodes/selva_lora_trainer.py
+1 -1
View File
@@ -139,7 +139,7 @@ def _eval_sample(generator, feature_utils_orig, dataset, seq_cfg, device, dtype,
elif audio.dim() == 3 and audio.shape[1] != 1: elif audio.dim() == 3 and audio.shape[1] != 1:
audio = audio.mean(dim=1, keepdim=True) audio = audio.mean(dim=1, keepdim=True)
target_rms = 10 ** (-23.0 / 20.0) # -23 dBFS matches training data target_rms = 10 ** (-27.0 / 20.0) # -27 dBFS matches measured RMS of training clips
rms = audio.pow(2).mean().sqrt().clamp(min=1e-8) rms = audio.pow(2).mean().sqrt().clamp(min=1e-8)
audio = audio * (target_rms / rms) audio = audio * (target_rms / rms)
peak = audio.abs().max().clamp(min=1e-8) peak = audio.abs().max().clamp(min=1e-8)
nodes/selva_sampler.py
+3 -3
View File
@@ -36,8 +36,8 @@ class SelvaSampler:
"tooltip": "Normalize output level. Uses RMS normalization to target_lufs rather than peak normalization, so level matches typical audio content.", "tooltip": "Normalize output level. Uses RMS normalization to target_lufs rather than peak normalization, so level matches typical audio content.",
}), }),
"target_lufs": ("FLOAT", { "target_lufs": ("FLOAT", {
"default": -20.0, "min": -40.0, "max": -6.0, "step": 1.0, "default": -27.0, "min": -40.0, "max": -6.0, "step": 1.0,
"tooltip": "Target RMS level in dBFS when normalize=True. -20 matches typical processed audio. Increase toward -14 for louder output, decrease toward -30 for quieter.", "tooltip": "Target RMS level in dBFS when normalize=True. -27 matches the measured RMS of LUFS-normalized training clips. Increase toward -20 for louder output.",
}), }),
}, },
} }
@@ -49,7 +49,7 @@ class SelvaSampler:
CATEGORY = SELVA_CATEGORY CATEGORY = SELVA_CATEGORY
DESCRIPTION = "Generates audio from video features using SelVA's flow matching ODE. Supports text prompts and negative prompts via classifier-free guidance." DESCRIPTION = "Generates audio from video features using SelVA's flow matching ODE. Supports text prompts and negative prompts via classifier-free guidance."
def generate(self, model, features, prompt, negative_prompt, duration, steps, cfg_strength, seed, normalize=True, target_lufs=-20.0): def generate(self, model, features, prompt, negative_prompt, duration, steps, cfg_strength, seed, normalize=True, target_lufs=-27.0):
import dataclasses import dataclasses
from selva_core.model.flow_matching import FlowMatching from selva_core.model.flow_matching import FlowMatching