Skip to content

Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.

wav is not exist

游雁0f7ce31c2a4e
diff --git a/funasr/datasets/sense_voice_datasets/datasets.py b/funasr/datasets/sense_voice_datasets/datasets.py
index dfadbab76..5d80956b1 100644
--- a/funasr/datasets/sense_voice_datasets/datasets.py
+++ b/funasr/datasets/sense_voice_datasets/datasets.py
@@ -2,7 +2,7 @@ import logging
import torch
import random
-
+import traceback
from funasr.register import tables
from funasr.utils.load_utils import extract_fbank, load_audio_text_image_video
@@ -73,15 +73,17 @@ class SenseVoiceDataset(torch.utils.data.Dataset):
if idx == 0:
index_cur = index
else:
- if index <= self.retry:
- index_cur = index + idx
- else:
- index_cur = torch.randint(0, index, ()).item()
+ index_cur = torch.randint(0, len(self.index_ds), ()).item()
item = self.index_ds[index_cur]
source = item["source"]
- data_src = load_audio_text_image_video(source, fs=self.fs)
+ try:
+ data_src = load_audio_text_image_video(source, fs=self.fs)
+ except Exception as e:
+ logging.error(f"Loading wav failed! {str(e)}, {traceback.format_exc()}")
+ continue
+
if self.preprocessor_speech:
data_src = self.preprocessor_speech(data_src, fs=self.fs)
speech, speech_lengths = extract_fbank(