| |
| """ |
| Audio processing utilities: |
| - Voice/instrumental separation (Demucs) |
| - Voice conversion (RVC) |
| - Audio mixing and synthesis |
| """ |
|
|
| from pathlib import Path |
| from pydub import AudioSegment |
| import subprocess |
| import sys |
| import os |
| import tempfile |
| import shutil |
| import traceback |
|
|
| |
| def load_audio(file_path): |
| """Load audio file""" |
| import librosa |
| audio, sr = librosa.load(file_path, sr=None) |
| return audio, sr |
|
|
| def save_audio(file_path, audio, sr): |
| """Save audio file""" |
| import soundfile as sf |
| sf.write(file_path, audio, sr) |
|
|
| |
| def separate_vocals_and_instrumental(input_audio: Path, output_dir: Path): |
| """ |
| Separate vocals and instrumental using Demucs |
| |
| Returns: |
| tuple: (vocals_path, instrumental_path) or (None, None) if failed |
| """ |
| output_dir = Path(output_dir) |
| output_dir.mkdir(parents=True, exist_ok=True) |
| |
| vocals_path = output_dir / "vocals.wav" |
| instrumental_path = output_dir / "instrumental.wav" |
| |
| try: |
| print("🎵 Starting vocal separation with Demucs...") |
| |
| cmd = [ |
| sys.executable, "-m", "demucs", |
| "--two-stems", "vocals", |
| "-n", "htdemucs", |
| "-o", str(output_dir), |
| str(input_audio) |
| ] |
| |
| result = subprocess.run( |
| cmd, |
| capture_output=True, |
| text=True, |
| timeout=600 |
| ) |
| |
| |
| input_stem = Path(input_audio).stem |
| demucs_output = output_dir / "htdemucs" / input_stem |
| |
| if demucs_output.exists(): |
| vocals_file = demucs_output / "vocals.wav" |
| no_vocals_file = demucs_output / "no_vocals.wav" |
| |
| if vocals_file.exists() and no_vocals_file.exists(): |
| shutil.move(str(vocals_file), str(vocals_path)) |
| shutil.move(str(no_vocals_file), str(instrumental_path)) |
| |
| |
| shutil.rmtree(output_dir / "htdemucs", ignore_errors=True) |
| |
| print(f"✅ Separation complete!") |
| return vocals_path, instrumental_path |
| |
| print(f"⚠️ Demucs output not found") |
| return None, None |
| |
| except subprocess.TimeoutExpired: |
| print("⚠️ Demucs timeout, skipping separation") |
| return None, None |
| except Exception as e: |
| print(f"⚠️ Separation failed: {e}") |
| return None, None |
|
|
| |
| def merge_vocals_and_instrumental(converted_vocals: Path, instrumental: Path, output_file: Path): |
| """ |
| Merge converted vocals with instrumental |
| |
| Returns: |
| bool: Success status |
| """ |
| try: |
| vocals_audio = AudioSegment.from_file(str(converted_vocals)) |
| instrumental_audio = AudioSegment.from_file(str(instrumental)) |
| |
| |
| target_sr = 44100 |
| if vocals_audio.frame_rate != target_sr: |
| vocals_audio = vocals_audio.set_frame_rate(target_sr) |
| if instrumental_audio.frame_rate != target_sr: |
| instrumental_audio = instrumental_audio.set_frame_rate(target_sr) |
| |
| |
| if vocals_audio.channels == 1: |
| vocals_audio = vocals_audio.set_channels(2) |
| if instrumental_audio.channels == 1: |
| instrumental_audio = instrumental_audio.set_channels(2) |
| |
| |
| vocals_len = len(vocals_audio) |
| instrumental_len = len(instrumental_audio) |
| |
| if vocals_len > instrumental_len: |
| vocals_audio = vocals_audio[:instrumental_len] |
| elif instrumental_len > vocals_len: |
| instrumental_audio = instrumental_audio[:vocals_len] |
| |
| |
| vocals_dBFS = vocals_audio.dBFS |
| instrumental_dBFS = instrumental_audio.dBFS |
| target_vocals_dBFS = instrumental_dBFS - 4.0 |
| |
| if vocals_dBFS > target_vocals_dBFS: |
| vocals_audio = vocals_audio + (target_vocals_dBFS - vocals_dBFS) |
| |
| |
| mixed_audio = instrumental_audio.overlay(vocals_audio) |
| |
| |
| mixed_dBFS = mixed_audio.dBFS |
| if mixed_dBFS < -25: |
| mixed_audio = mixed_audio + (-20.0 - mixed_dBFS) |
| elif mixed_dBFS > -10: |
| mixed_audio = mixed_audio + (-20.0 - mixed_dBFS) |
| |
| |
| output_file = Path(output_file) |
| output_file.parent.mkdir(parents=True, exist_ok=True) |
| mixed_audio.export(str(output_file), format="wav") |
| |
| return True |
| |
| except Exception as e: |
| print(f"❌ Merge failed: {e}") |
| traceback.print_exc() |
| return False |
|
|
| def optimize_audio(input_file: Path, output_file: Path): |
| """Optimize audio quality""" |
| try: |
| audio = AudioSegment.from_file(str(input_file)) |
| |
| if audio.frame_rate != 44100: |
| audio = audio.set_frame_rate(44100) |
| if audio.channels == 1: |
| audio = audio.set_channels(2) |
| |
| |
| dBFS = audio.dBFS |
| if dBFS < -30 or dBFS > -10: |
| audio = audio + (-20.0 - dBFS) |
| |
| output_file = Path(output_file) |
| output_file.parent.mkdir(parents=True, exist_ok=True) |
| audio.export(str(output_file), format="wav") |
| |
| return True |
| except Exception as e: |
| print(f"Audio optimization failed: {e}") |
| shutil.copy2(input_file, output_file) |
| return False |
|
|
| |
| def simple_mix_audio(voice_file: Path, background_music: Path, output_file: Path, |
| voice_volume_db: float = 0.0, music_volume_db: float = -10.0): |
| """ |
| Simple mix mode: overlay voice on background music |
| """ |
| try: |
| voice_audio = AudioSegment.from_file(str(voice_file)) |
| music_audio = AudioSegment.from_file(str(background_music)) |
| |
| if voice_volume_db != 0.0: |
| voice_audio = voice_audio + voice_volume_db |
| if music_volume_db != 0.0: |
| music_audio = music_audio + music_volume_db |
| |
| |
| voice_len = len(voice_audio) |
| music_len = len(music_audio) |
| |
| if voice_len > music_len: |
| voice_audio = voice_audio[:music_len] |
| elif music_len > voice_len: |
| num_loops = (music_len // voice_len) + 1 |
| voice_audio = voice_audio * num_loops |
| voice_audio = voice_audio[:music_len] |
| |
| mixed_audio = music_audio.overlay(voice_audio) |
| |
| output_file = Path(output_file) |
| output_file.parent.mkdir(parents=True, exist_ok=True) |
| mixed_audio.export(str(output_file), format="wav") |
| |
| return True |
| |
| except Exception as e: |
| print(f"Simple mix failed: {e}") |
| return False |
