Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
Python asraudiochineseemotion-recognitionfunasrmcp-servermultilingual-asropenai-compatible-apiparaformerpunctuationpytorchreal-time-asrspeaker-diarizationspeech-recognitionspeech-to-textstreaming-asrtranscriptionvllmvoice-activity-detectionwhisper-alternative
fix bug
游雁56986acaa78b
diff --git a/funasr/datasets/audio_datasets/samplers.py b/funasr/datasets/audio_datasets/samplers.pyindex bddf18638..f8f744c55 100644--- a/funasr/datasets/audio_datasets/samplers.py+++ b/funasr/datasets/audio_datasets/samplers.py@@ -392,7 +392,7 @@ class CustomDistributedBufferDynamicBatchSampler(DistributedSampler):)batch = []max_len_in_batch = 0- count = 0+ count = 1for idx in buffer:original_sample_length = self.dataset.get_source_len(idx)if original_sample_length > self.max_token_length:@@ -410,7 +410,7 @@ class CustomDistributedBufferDynamicBatchSampler(DistributedSampler):buffer_batches.append(batch)batch = [idx]max_len_in_batch = sample_length- count = 0+ count = 1if batch:buffer_batches.append(batch)