test

app.py

@@ -5,6 +5,7 @@ import sys
 sys.path.append('.')
 import gradio as gr
 os.system('pip install -U torchtext==0.8.0')
+os.system('pip install pydub')  # 添加对pydub库的依赖
 #os.system('python setup.py install --install-dir .')
 from scipy.io import wavfile
 
@@ -14,6 +15,11 @@ os.system('chmod a+x ./scripts/*/*.sh')
 os.system('./separate_scripts/download_checkpoints.sh')
 
 def inference(audio):
+    # convert mp3 to wav if needed
+    if audio.name.endswith(".mp3"):
+        audio_wav = audio.name[:-4] + ".wav"
+        AudioSegment.from_file(audio.name).export(audio_wav, format="wav")
+        audio.name = audio_wav
     # read the file and get the sample rate and data
     rate, data = wavfile.read(audio.name)