update

app.py

@@ -78,7 +78,7 @@ def inference_model(audio_input, model):
         y_g_hat = model(spec_gt)
 
         audio_gen = y_g_hat.squeeze().cpu()
-        spec_gen = get_mel_spectrogram(audio_gen.unsqueeze(0))
+        spec_gen = get_mel_spectrogram(audio_gen.unsqueeze(0), model.h)
         audio_gen = audio_gen.numpy()  # [T], float [-1, 1]
         audio_gen = (audio_gen * MAX_WAV_VALUE).astype("int16")  # [T], int16
         spec_gen = spec_gen.squeeze().numpy()  # [C, T_frame]