Update app.py

app.py

@@ -13,23 +13,22 @@ asr_pipe = pipeline("automatic-speech-recognition", model="openai/whisper-base",
 
 
 
-model = VitsModel.from_pretrained("facebook/mms-tts-hin")
-processor = VitsTokenizer.from_pretrained("facebook/mms-tts-hin")
+model = VitsModel.from_pretrained("facebook/mms-tts-spa")
+processor = VitsTokenizer.from_pretrained("facebook/mms-tts-spa")
 
 
 
 
 def translate(audio):
-    outputs = asr_pipe(audio, max_new_tokens=256, generate_kwargs={"task": "translate", "language": "es"})
+    outputs = asr_pipe(audio, max_new_tokens=256, generate_kwargs={"language": "es","task": "transcribe"})
     return outputs["text"]
 
 
 def synthesise(text):
     inputs = processor(text=text, return_tensors="pt")
     with torch.no_grad():
-        speech = model(inputs["input_ids"].to(device))
-
-    return speech
+      output = model(**inputs)
+    return output['audio']
 
 
 def speech_to_speech_translation(audio):