Update app.py

app.py

@@ -15,8 +15,8 @@ asr_pipe = pipeline("automatic-speech-recognition", model="openai/whisper-base",
 # load text-to-speech checkpoint and speaker embeddings
 processor = SpeechT5Processor.from_pretrained("microsoft/speecht5_tts")
 
-model = SpeechT5ForTextToSpeech.from_pretrained("microsoft/speecht5_tts").to(device)
-vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan").to(device)
+model = SpeechT5ForTextToSpeech.from_pretrained("microsoft/speecht5_tts").to("cuda")
+vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan").to("cuda")
 
 embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
 speaker_embeddings = torch.tensor(embeddings_dataset[7306]["xvector"]).unsqueeze(0)
@@ -29,7 +29,7 @@ def translate(audio):
 @spaces.GPU
 def synthesise(text):
     inputs = processor(text=text, return_tensors="pt")
-    speech = model.generate_speech(inputs["input_ids"].to(device), speaker_embeddings.to(device), vocoder=vocoder)
+    speech = model.generate_speech(inputs["input_ids"].to("cuda"), speaker_embeddings.to("cuda"), vocoder=vocoder)
     return speech.cpu()
 
 @spaces.GPU