Update app.py

app.py

@@ -4,7 +4,7 @@ import soundfile as sf
 import uuid
 import datetime
 import shutil
-from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
+from transformers import pipeline
 import scipy.io.wavfile
 import numpy as np
 
@@ -71,13 +71,9 @@ def tts_tibetan(input_text):
         # Perform TTS inference for each sentence
         speech = synthesiser(sentence)
         
-        # Extract the audio data and sampling rate from the pipeline output
-        audio_data = np.array(speech["audio"])
-        sample_rate = speech["sampling_rate"]
-
         # Save each sentence as a separate WAV file
         wav_path = f"{user_dir}/s_{str(i).zfill(10)}.wav"
-        scipy.io.wavfile.write(wav_path, rate=sample_rate, data=audio_data.astype(np.int16))  # Ensure correct format
+        scipy.io.wavfile.write(wav_path, rate=speech["sampling_rate"], data=speech["audio"][0])
 
     # Combine the generated audio into one file
     combined_file_path = combine_wav(user_dir, timestamp)