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
decoding
游雁f97f3e8dd5b4
diff --git a/funasr/bin/train_ds.py b/funasr/bin/train_ds.pyindex 41ecbe4bc..da99adccc 100644--- a/funasr/bin/train_ds.py+++ b/funasr/bin/train_ds.py@@ -66,7 +66,6 @@ def main(**kwargs):# open tf32torch.backends.cuda.matmul.allow_tf32 = kwargs.get("enable_tf32", True)- rank = int(os.environ.get("RANK", 0))local_rank = int(os.environ.get("LOCAL_RANK", 0))world_size = int(os.environ.get("WORLD_SIZE", 1))@@ -84,6 +83,8 @@ def main(**kwargs):dist.init_process_group(backend=kwargs.get("backend", "nccl"), init_method="env://")torch.cuda.set_device(local_rank)+ rank = dist.get_rank()+logging.info("Build model, frontend, tokenizer")device = kwargs.get("device", "cuda")kwargs["device"] = "cpu"