Update app.py

app.py

@@ -4,6 +4,8 @@ import numpy as np
 import torch
 import os
 import torch
+from scipy.io import wavfile
+import scipy.signal as sps
 from speechbrain.pretrained import EncoderClassifier
 from transformers import AutoProcessor, AutoModelForTextToSpectrogram, SpeechT5HifiGan
 
@@ -29,10 +31,13 @@ def create_speaker_embedding(waveform):
 
 def prepare_data(temp_text, temp_audio):
     rate, audio_data = temp_audio
+    new_rate = 16000
+    number_of_samples = round(len(audio_data) * float(new_rate) / rate)
+    audio_data = sps.resample(audio_data, number_of_samples)
     example = processor(
         text=temp_text,
         audio_target=audio_data,
-        sampling_rate=rate,
+        sampling_rate=16000,
         return_attention_mask=False,)
     example["speaker_embeddings"] = create_speaker_embedding(audio_data)
     example_embeddings = torch.tensor(example["speaker_embeddings"]).unsqueeze(0)