Update app.py

app.py

@@ -5,6 +5,7 @@ import shutil
 #from huggingface_hub import snapshot_download
 import numpy as np
 from scipy.io import wavfile
+from pydub import AudioSegment
 """
 model_ids = [
     'suno/bark',
@@ -25,12 +26,40 @@ model.load_checkpoint(config, checkpoint_dir="checkpoints/bark", eval=True)
 from TTS.api import TTS
 tts = TTS("tts_models/multilingual/multi-dataset/bark", gpu=True)
 
+def cut_wav(input_path, max_duration):
+    # Load the WAV file
+    audio = AudioSegment.from_wav(input_path)
+    
+    # Calculate the duration of the audio
+    audio_duration = len(audio) / 1000  # Convert milliseconds to seconds
+    
+    # Determine the duration to cut (maximum of max_duration and actual audio duration)
+    cut_duration = min(max_duration, audio_duration)
+    
+    # Cut the audio
+    cut_audio = audio[:int(cut_duration * 1000)]  # Convert seconds to milliseconds
+    
+    # Get the input file name without extension
+    file_name = os.path.splitext(os.path.basename(input_path))[0]
+    
+    # Construct the output file path with the original file name and "_cut" suffix
+    output_path = f"{file_name}_cut.wav"
+    
+    # Save the cut audio as a new WAV file
+    cut_audio.export(output_path, format="wav")
+
+    return output_path
+    
 def infer(prompt, input_wav_file):
 
+    print("CUT AUDIO FILE TO MAX 20 SECONDS")
+
+    cut_input_wav_file = cut_wav(input_wav_file, max_duration=20)
+
     print("SAVING THE AUDIO FILE TO WHERE IT BELONGS")
 
     # Path to your WAV file
-    source_path = input_wav_file
+    source_path = cut_input_wav_file
 
     # Destination directory
     destination_directory = "bark_voices"