LRC Generator

An automatic lyrics timing generator based on OpenAI Whisper. This tool automatically aligns lyrics with audio files to generate timestamped LRC files.

Features

• Supports multiple audio formats (MP3, WAV, etc.)
• Uses Whisper for accurate speech recognition
• Automatic lyrics-to-audio alignment
• Generates standard LRC format files
• Supports special characters in lyrics
• Handles repeated lyrics sections
• Provides detailed debugging information

Requirements

• Python 3.8 or higher
• FFmpeg (for audio processing)
• PyTorch (for Whisper model)

Installation

1. Install FFmpeg (if not already installed):

   # macOS
   brew install ffmpeg
   
   # Ubuntu/Debian
   sudo apt-get install ffmpeg

2. Clone the repository:

   git clone [repository-url]
   cd LRC-Generator

3. Install dependencies:

   pip install -r requirements.txt

Usage

Basic usage:

lrc-gen generate --audio "path/to/audio" --lyrics "path/to/lyrics"

Example:

lrc-gen generate --audio "./songs/my_song.mp3" --lyrics "./lyrics/my_lyrics.txt"

Handling Filenames with Spaces or Special Characters

If your filenames contain spaces or special characters, you can:

1. Use quotes:

   lrc-gen generate --audio "./songs/my song.mp3" --lyrics "./lyrics/my lyrics.txt"

2. Use backslash escaping:

   lrc-gen generate --audio ./songs/my\ song.mp3 --lyrics ./lyrics/my\ lyrics.txt

File Format Requirements

Audio Files

• Supported formats: MP3, WAV, M4A, FLAC, etc.
• High-quality audio recommended for better recognition

Lyrics Files

• Plain text file (.txt)
• UTF-8 encoding
• One line per lyric
• No timestamps
• Arranged in singing order

Example lyrics file format:

verse 1
First line of lyrics
Second line of lyrics
chorus
This is the chorus
Second line of chorus

Output Files

• Generated LRC file will have the same name as the lyrics file (different extension)
• Location: Same directory as the lyrics file
• Format: Standard LRC format, UTF-8 encoding (with BOM)

Example output:

[ti:Song Title]
[ar:Artist]
[al:Album]
[length:03:45]

[00:01.23]First line of lyrics
[00:05.67]Second line of lyrics
[00:10.89]This is the chorus

Debug Information

The program generates two debug files:

1. whisper_transcription.txt: Contains Whisper's speech recognition results
2. debug_segments.txt: Contains detailed matching process information

Important Notes

1. First run will download the Whisper model, requiring internet connection
2. Processing time depends on audio length and chosen model size
3. High-quality audio files recommended for better recognition
4. If lyrics matching is not ideal:
   • Check lyrics text accuracy
   • Ensure correct lyrics order
   • Review debug files for detailed matching process

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contributing

Issues and Pull Requests are welcome!