Skip to content

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

感觉应该从文件读取更合适,因为上面判断了文件存在,且可以读取,如果本身是文本的话,下面也会有逻辑进行处理 (#2452)

天地e24dbdc496de
diff --git a/funasr/utils/load_utils.py b/funasr/utils/load_utils.py
index 072b578ef..c38f1d937 100644
--- a/funasr/utils/load_utils.py
+++ b/funasr/utils/load_utils.py
@@ -107,7 +107,8 @@ def load_audio_text_image_video(
data_or_path_or_list
).squeeze() # [n_samples,]
elif data_type == "text" and tokenizer is not None:
- data_or_path_or_list = tokenizer.encode(data_or_path_or_list)
+ with open(data_or_path_or_list, "r") as f:
+ data_or_path_or_list = tokenizer.encode(f.read().strip())
elif data_type == "image": # undo
pass
elif data_type == "video": # undo