updated

app.py

@@ -0,0 +1,129 @@
+import gradio as gr
+import numpy as np
+import torch
+import os
+import time
+from scipy.io import wavfile
+
+# Explicitly import Bark components
+from bark import generate_audio, SAMPLE_RATE
+from bark.generation import preload_models, load_model, generate_text_semantic
+
+class VoiceCloningApp:
+    def __init__(self):
+        # Create working directory
+        self.base_dir = os.path.dirname(os.path.abspath(__file__))
+        self.working_dir = os.path.join(self.base_dir, "working_files")
+        os.makedirs(self.working_dir, exist_ok=True)
+        
+        # Explicit model loading with error handling
+        try:
+            print("Attempting to load Bark models...")
+            preload_models()
+            print("Bark models loaded successfully.")
+        except Exception as e:
+            print(f"Error loading Bark models: {e}")
+            raise RuntimeError(f"Could not load Bark models: {e}")
+
+    def process_reference_audio(self, audio_data):
+        """Simple audio processing"""
+        if audio_data is None:
+            return "Please provide an audio input"
+        
+        try:
+            # Unpack audio data
+            sample_rate, audio_array = audio_data
+            
+            # Normalize audio
+            audio_array = audio_array / np.max(np.abs(audio_array))
+            
+            # Save reference audio
+            filename = f"reference_{int(time.time())}.wav"
+            filepath = os.path.join(self.working_dir, filename)
+            wavfile.write(filepath, sample_rate, audio_array)
+            
+            return "✅ Audio captured successfully!"
+        
+        except Exception as e:
+            return f"Error processing audio: {str(e)}"
+
+    def generate_speech(self, text):
+        """Generate speech using Bark"""
+        if not text or not text.strip():
+            return None, "Please enter some text to speak"
+        
+        try:
+            # Generate audio with explicit error handling
+            print(f"Generating speech for text: {text}")
+            
+            # Use alternative generation method
+            semantic_tokens = generate_text_semantic(
+                text,
+                history_prompt=None,
+                temp=0.7,
+                min_eos_p=0.05,
+            )
+            
+            # Generate audio from semantic tokens
+            audio_array = generate_audio(
+                semantic_tokens,
+                history_prompt=None,
+                temp=0.7
+            )
+            
+            # Save generated audio
+            filename = f"generated_speech_{int(time.time())}.wav"
+            filepath = os.path.join(self.working_dir, filename)
+            wavfile.write(filepath, SAMPLE_RATE, audio_array)
+            
+            return filepath, None
+        
+        except Exception as e:
+            print(f"Speech generation error: {e}")
+            return None, f"Error generating speech: {str(e)}"
+
+def create_interface():
+    app = VoiceCloningApp()
+    
+    # Use the most basic Gradio theme to avoid font issues
+    with gr.Blocks() as interface:
+        gr.Markdown("# 🎙️ Voice Cloning App")
+        
+        with gr.Row():
+            with gr.Column():
+                gr.Markdown("## 1. Capture Reference Voice")
+                reference_audio = gr.Audio(sources=["microphone", "upload"], type="numpy")
+                process_btn = gr.Button("Process Reference Voice")
+                process_output = gr.Textbox(label="Processing Result")
+                
+            with gr.Column():
+                gr.Markdown("## 2. Generate Speech")
+                text_input = gr.Textbox(label="Enter Text to Speak")
+                generate_btn = gr.Button("Generate Speech")
+                audio_output = gr.Audio(label="Generated Speech")
+                error_output = gr.Textbox(label="Errors", visible=True)
+        
+        # Bind functions
+        process_btn.click(
+            fn=app.process_reference_audio, 
+            inputs=reference_audio, 
+            outputs=process_output
+        )
+        
+        generate_btn.click(
+            fn=app.generate_speech, 
+            inputs=text_input, 
+            outputs=[audio_output, error_output]
+        )
+    
+    return interface
+
+if __name__ == "__main__":
+    interface = create_interface()
+    interface.launch(
+        share=False,
+        debug=True,
+        show_error=True,
+        server_name='0.0.0.0',
+        server_port=7860
+    )

requirements.txt

@@ -0,0 +1,7 @@
+gradio>=3.50.2
+numpy
+scipy
+soundfile
+torch
+transformers
+git+https://github.com/suno-ai/bark.git