Update voice_cloner.py

voice_cloner.py

@@ -1,18 +1,30 @@
-from transformers import BarkModel, BarkProcessor
+# voice_cloner.py
+from transformers import BarkModel, AutoProcessor
+import torchaudio
 import torch
+import os
 
-# You could switch to xtts-v2 or Tortoise if needed
-def clone_and_generate_text(text, reference_audio_path, output_path="clone_output.wav"):
-    processor = BarkProcessor.from_pretrained("suno/bark")
+def clone_and_generate_text(text, reference_audio_path, language="English", emotion="Neutral"):
+    processor = AutoProcessor.from_pretrained("suno/bark")
     model = BarkModel.from_pretrained("suno/bark")
 
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+    model.to(device)
 
-    # Preprocess input
-    inputs = processor(text=text, voice_preset="v2/en_speaker_9", return_tensors="pt").to("cpu")
+    # Load and process reference audio
+    speech_array, sampling_rate = torchaudio.load(reference_audio_path)
+    speech_array = torchaudio.functional.resample(speech_array, sampling_rate, 16000)
+    speech_array = speech_array.mean(dim=0).unsqueeze(0)  # mono
 
-    # Generate speech
-    speech = model.generate(**inputs)
+    inputs = processor(
+        text=text,
+        voice_preset="v2/en_speaker_9",  # generic fallback voice
+        return_tensors="pt"
+    ).to(device)
 
-    # Save output
+    with torch.no_grad():
+        speech = model.generate(**inputs)
+
+    output_path = "output_voice.wav"
     torchaudio.save(output_path, speech.cpu(), 22050)
     return output_path