Update app.py

app.py

@@ -1,6 +1,6 @@
 import torch
 import os
-from transformers import pipeline, VitsModel, VitsTokenizer
+from transformers import pipeline, VitsModel, VitsTokenizer, SpeechT5ForTextToSpeech, SpeechT5HifiGan, SpeechT5Processor
 import numpy as np
 os.system("pip install git+https://github.com/openai/whisper.git")
 import gradio as gr
@@ -43,7 +43,7 @@ tokenizer = VitsTokenizer.from_pretrained("facebook/mms-tts-eng")
 def translate(audio):
     # return inference(audio)
     outputs = pipe(audio, max_new_tokens=256,
-                   generate_kwargs={"task": "translate", "language": "swedish"})
+                   generate_kwargs={"task": "translate"})
     return outputs["text"]
 
 
@@ -70,7 +70,7 @@ def predict(transType, language, audio, audio_mic = None):
         print("debug1:", audio,"debug2", audio_mic)
         if not audio and audio_mic:
             audio = audio_mic
-        audio = audio[1]
+        
         if transType == "Text":
             return translate(audio), None
         if transType == "Audio":
@@ -95,8 +95,8 @@ demo = gr.Interface(
     inputs=[
         gr.Radio(label="Choose your output format", choices=transTypes),
         gr.Radio(label="Choose a source language", choices=supportLangs, value="Swedish"),
-        #gr.Audio(label="Import an audio", sources="upload", type="filepath"),
-        gr.Audio(label="Import an audio", sources="upload", type="numpy"),
+        gr.Audio(label="Import an audio", sources="upload", type="filepath"),
+        #gr.Audio(label="Import an audio", sources="upload", type="numpy"),
         gr.Audio(label="Record an audio", sources="microphone", type="filepath"),
     ],
     outputs=[