feat: whisper使用开源模型

main.py

@@ -102,22 +102,22 @@ def predict(input, history=[]):
         return responses, video_html, history
 
 
-# 通过openai whisper 语音识别
-# def transcribe(audio):
-#     os.rename(audio, audio + '.wav')
-#     audio_file = open(audio + '.wav', "rb")
-#     transcript = openai.Audio.transcribe(
-#         "whisper-1", audio_file, prompt="这是一段简体中文的问题。")
-#     return transcript['text']
-
-
 # 通过openai whisper 语音识别
 def transcribe(audio):
     os.rename(audio, audio + '.wav')
     audio_file = open(audio + '.wav', "rb")
-    transcriber = pipeline(model="openai/whisper-medium", device=0)
-    result = transcriber(audio_file)
-    return result['text']
+    transcript = openai.Audio.transcribe(
+        "whisper-1", audio_file, prompt="这是一段简体中文的问题。")
+    return transcript['text']
+
+
+# 通过openai whisper 语音识别
+# def transcribe(audio):
+#     os.rename(audio, audio + '.wav')
+#     audio_file = open(audio + '.wav', "rb")
+#     transcriber = pipeline(model="openai/whisper-medium", device=0)
+#     result = transcriber(audio_file)
+#     return result['text']
 
 
 def process_audio(audio, history=[]):