Update app.py

app.py

@@ -30,11 +30,12 @@ def text_to_audio(text):
     mel = mel_output.detach().cpu().numpy().astype(np.float32)  # (80, frames)
     
     # 2. Preparar para generator
-    mel_input = np.expand_dims(mel,axis=-1) # (1, 80, frames, 1)
+    mel_input = np.expand_dims(mel,axis=0) # (1, 80, frames, 1)
+    mel_input = np.expand_dims(mel_input,axis=-1) # (1, 80, frames, 1)
 
     # 3. Usar generator para generar audio
     fake_audio = generator(mel_input, training=False)
-    fake_audio = tf.squeeze(fake_audio, axis=0).numpy()  # (samples,)
+    fake_audio = np.squeeze(fake_audio, axis=-1)  # (samples,)
 
     # 4. Asegurar que esté en [-1, 1] para audio
     fake_audio = np.clip(fake_audio, -1.0, 1.0)