asr

requirements.txt

@@ -2,4 +2,6 @@ gradio==2.9.1
 torch
 torchaudio
 transformers
-numpy
+numpy
+scipy==1.13.1
+umsc

tts.py

@@ -17,7 +17,7 @@ def synthesize(text, model_id):
     inputs = processor(text, return_tensors="pt").to(device)
 
     with torch.no_grad():
-        output = tts_model(**inputs).waveform.cpu()  # Move output back to CPU for saving
+        output = model(**inputs).waveform.cpu()  # Move output back to CPU for saving
     
     output_path = "tts_output.wav"
     sample_rate = 16000