Create runfile.py

runfile.py

@@ -0,0 +1,67 @@
+import librosa
+import numpy as np
+import soundfile as sf
+import gradio as gr
+import argparse
+
+def auto_tune(audio_path, output_path=None, target_pitch=440.0):
+    # Load the audio file
+    y, sr = librosa.load(audio_path)
+    
+    # Extract pitch using librosa's piptrack method
+    pitches, magnitudes = librosa.core.piptrack(y=y, sr=sr)
+    
+    # Select pitches with higher magnitudes
+    pitches = [p for p, m in zip(pitches, magnitudes) if m > np.median(magnitudes)]
+    
+    # Remove outliers
+    pitches = np.array(pitches)
+    pitches = pitches[pitches > 0]  # Keep only positive values
+    
+    # Calculate the tuning ratio to shift the pitch to the target pitch
+    median_pitch = np.median(pitches)
+    tuning_ratio = target_pitch / median_pitch
+    
+    # Adjust the pitch of the audio
+    y_tuned = librosa.effects.pitch_shift(y, sr, n_steps=np.log2(tuning_ratio))
+    
+    # Save the tuned audio if output path is specified
+    if output_path:
+        sf.write(output_path, y_tuned, sr)
+    
+    return y_tuned, sr
+
+def run_cli():
+    parser = argparse.ArgumentParser(description="Auto-tune a voice recording.")
+    parser.add_argument("input", type=str, help="Path to the input audio file")
+    parser.add_argument("output", type=str, help="Path to save the tuned audio file")
+    parser.add_argument("--target_pitch", type=float, default=440.0, help="Target pitch (default: 440 Hz)")
+    
+    args = parser.parse_args()
+    y_tuned, sr = auto_tune(args.input, args.output, args.target_pitch)
+    print(f"Saved tuned audio to {args.output}")
+
+def run_gradio():
+    def gradio_interface(audio, target_pitch):
+        audio_path = "input_audio.wav"
+        output_path = "output_tuned.wav"
+        sf.write(audio_path, audio[1], audio[0])
+        y_tuned, sr = auto_tune(audio_path, target_pitch=target_pitch)
+        return (sr, y_tuned)
+
+    iface = gr.Interface(
+        fn=gradio_interface,
+        inputs=[gr.Audio(source="microphone", type="numpy"), gr.Number(value=440.0, label="Target Pitch")],
+        outputs=gr.Audio(type="numpy", label="Tuned Audio"),
+        title="Auto-Tune Voice",
+        description="Upload or record your voice and apply auto-tune.",
+    )
+    iface.launch()
+
+if __name__ == "__main__":
+    # Check if running in CLI or as a Gradio app
+    import sys
+    if len(sys.argv) > 1:
+        run_cli()
+    else:
+        run_gradio()