Fix app

app.py

@@ -18,9 +18,9 @@ class GradioInference():
       self.yt = YouTube(link)
     path = self.yt.streams.filter(only_audio=True)[0].download(filename="tmp.mp4")
     subprocess.run([
-        'ffmpeg', '-i', 'tmp.mp4', 'tmp.mp3'
+        'ffmpeg', '-i', 'tmp.mp4','-vn', '-acodec', 'pcm_s16le', '-ac', '1', '-ar', '44100', '-f' 'wav','tmp.wav'
     ])
-    sound_data = soundfile.read('tmp.mp3')
+    sound_data = soundfile.read('tmp.wav')
     input_features = self.processor(sound_data, return_tensors="pt").input_features
     forced_decoder_ids = self.processor.get_decoder_prompt_ids(language = "sv", task = "translate")
     predicted_ids = self.model.generate(input_features, forced_decoder_ids = forced_decoder_ids)