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 device bug
shixian.shic47ad73e46bb
diff --git a/funasr/auto/auto_model.py b/funasr/auto/auto_model.pyindex d07221976..3986a110c 100644--- a/funasr/auto/auto_model.py+++ b/funasr/auto/auto_model.py@@ -134,8 +134,6 @@ class AutoModel:self.spk_model = spk_modelself.spk_kwargs = spk_kwargsself.model_path = kwargs.get("model_path")--def build_model(self, **kwargs):assert "model" in kwargs@@ -146,7 +144,7 @@ class AutoModel:set_all_random_seed(kwargs.get("seed", 0))device = kwargs.get("device", "cuda")- if not torch.cuda.is_available() or kwargs.get("ngpu", 0) == 0:+ if not torch.cuda.is_available() or kwargs.get("ngpu", 1) == 0:device = "cpu"kwargs["batch_size"] = 1kwargs["device"] = device@@ -200,8 +198,6 @@ class AutoModel:res = self.model(*args, kwargs)return res--def generate(self, input, input_len=None, **cfg):if self.vad_model is None:return self.inference(input, input_len=input_len, **cfg)