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
bugfix for paraformer-streaming
游雁a28de72b1710
diff --git a/funasr/utils/load_utils.py b/funasr/utils/load_utils.pyindex faafc0fca..37b3dbfbf 100644--- a/funasr/utils/load_utils.py+++ b/funasr/utils/load_utils.py@@ -76,7 +76,7 @@ def load_audio_text_image_video(for audio in data_or_path_or_list]if isinstance(data_or_path_or_list, str) and data_or_path_or_list.startswith(- ("http://", "https://")+ ("http://", "https://")): # download url to local filedata_or_path_or_list = download_from_url(data_or_path_or_list)@@ -112,7 +112,7 @@ def load_audio_text_image_video(elif isinstance(data_or_path_or_list, str) and data_type == "text" and tokenizer is not None:data_or_path_or_list = tokenizer.encode(data_or_path_or_list)elif isinstance(data_or_path_or_list, np.ndarray): # audio sample point- data_or_path_or_list = torch.from_numpy(data_or_path_or_list).squeeze() # [n_samples,]+ data_or_path_or_list = torch.from_numpy(data_or_path_or_list) # .squeeze() # [n_samples,]elif isinstance(data_or_path_or_list, str) and data_type == "kaldi_ark":data_mat = kaldiio.load_mat(data_or_path_or_list)if isinstance(data_mat, tuple):