wangbo/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2026-06-26 09:49:26 +08:00
Code Issues Actions 25 Packages Projects Releases Wiki Activity

Fix SaveAudioMP3 avcodec_open2 EINVAL crashes

Browse Source 
Two separate bugs in AudioSaveHelper.save_audio caused libmp3lame's
avcodec_open2 to fail with EINVAL ([Errno 22] Invalid argument):

1. codec_context.qscale is exposed by PyAV as a bool flag rather than
   the int quality scale libmp3lame expects, so assigning a level (1)
   to it for "V0" quality made avcodec_open2 reject the codec context.
   libmp3lame's default VBR mode is roughly equivalent to V0 quality,
   so the assignment is simply removed.

2. libmp3lame only accepts a fixed set of sample rates (8000, 11025,
   12000, 16000, 22050, 24000, 32000, 44100, 48000 Hz). Audio sources
   that output other rates (e.g. 40000 Hz from some RVC models) were
   passed through unchanged, causing avcodec_open2 to fail. MP3 output
   now remaps to the closest supported rate and resamples, mirroring
   the existing Opus rate-handling logic.

Verified by reproducing both failures directly against PyAV with the
unpatched code and confirming the patched AudioSaveHelper.save_audio
encodes successfully for both the "V0" quality and 40 kHz cases.

Fixes #13937
This commit is contained in:
nahcmon 2026-06-08 18:30:35 +02:00
parent 38f750d80e
commit 39f62a0d4c
1 changed files with 13 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
comfy_api/latest/_ui.py
View File

@ -260,6 +260,7 @@ class ImageSaveHelper:
class AudioSaveHelper:
"""A helper class with static methods to handle audio saving and metadata."""
_OPUS_RATES = [8000, 12000, 16000, 24000, 48000]
_MP3_RATES = [8000, 11025, 12000, 16000, 22050, 24000, 32000, 44100, 48000]
@staticmethod
def save_audio(
@ -303,12 +304,15 @@ class AudioSaveHelper:
break
if sample_rate not in AudioSaveHelper._OPUS_RATES: # Fallback if still not supported
sample_rate = 48000
elif format == "mp3" and sample_rate not in AudioSaveHelper._MP3_RATES:
# libmp3lame only accepts a fixed set of sample rates; remap to the closest one
sample_rate = min(AudioSaveHelper._MP3_RATES, key=lambda rate: abs(rate - sample_rate))
# Resample if necessary
if sample_rate != audio["sample_rate"]:
if not TORCH_AUDIO_AVAILABLE:
raise Exception("torchaudio is not available; cannot resample audio.")
waveform = torchaudio.functional.resample(waveform, audio["sample_rate"], sample_rate)
# Resample if necessary
if sample_rate != audio["sample_rate"]:
if not TORCH_AUDIO_AVAILABLE:
raise Exception("torchaudio is not available; cannot resample audio.")
waveform = torchaudio.functional.resample(waveform, audio["sample_rate"], sample_rate)
# Create output with specified format
output_buffer = BytesIO()
@ -335,8 +339,10 @@ class AudioSaveHelper:
elif format == "mp3":
out_stream = output_container.add_stream("libmp3lame", rate=sample_rate, layout=layout)
if quality == "V0":
# TODO i would really love to support V3 and V5 but there doesn't seem to be a way to set the qscale level, the property below is a bool
out_stream.codec_context.qscale = 1
# PyAV exposes codec_context.qscale as a bool flag, not the int quality
# scale libmp3lame expects, so assigning a level to it raises avcodec_open2
# EINVAL. libmp3lame's default VBR mode is roughly equivalent to V0 quality.
pass
elif quality == "128k":
out_stream.bit_rate = 128000
elif quality == "320k":