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

fix: fall back to soundfile for torchaudio.save when torchcodec unavailable

Browse Source 
Same torchcodec/FFmpeg issue as the load path, now on the eval sample save.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Ethanfel
2026-04-05 22:44:04 +02:00
parent afb3242eca
commit 004ea63f62
1 changed files with 5 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
nodes/selva_lora_trainer.py
+4
View File
@@ -508,7 +508,11 @@ class SelvaLoraTrainer:
dataset, seq_cfg, device, dtype) dataset, seq_cfg, device, dtype)
if wav is not None: if wav is not None:
wav_path = output_dir / f"sample_step{step:05d}.wav" wav_path = output_dir / f"sample_step{step:05d}.wav"
try:
torchaudio.save(str(wav_path), wav, sr) torchaudio.save(str(wav_path), wav, sr)
except RuntimeError:
import soundfile as sf
sf.write(str(wav_path), wav.squeeze(0).numpy(), sr)
print(f"[LoRA Trainer] Sample saved: {wav_path}", flush=True) print(f"[LoRA Trainer] Sample saved: {wav_path}", flush=True)
pbar_train.update(1) pbar_train.update(1)