changed to tacatron2

app.py

@@ -41,12 +41,12 @@ def process_text(text):
     return '\n'.join(processed)
 
 def text_to_speech(text):
-    """Convert text to speech using Hugging Face transformers."""
+    """Convert text to speech using Hugging Face's SpeechBrain model."""
     try:
-        tts_pipeline = pipeline("text-to-speech", model="espnet/kan-bayashi_ljspeech_vits", device=0 if torch.cuda.is_available() else -1)
+        tts_pipeline = pipeline("text-to-speech", model="speechbrain/tts-tacotron2-ljspeech", device=0 if torch.cuda.is_available() else -1)
         output = tts_pipeline(text)
 
-        audio_array = np.array(output["array"])
+        audio_array = np.array(output["audio"])
         sample_rate = output["sampling_rate"]
 
         audio_file = "output.wav"