Aligning lyrics to audio is the task of determining the precise time in a song when each individual word is being sung, i.e. the start and end time of every lyric. In the past, this timing information had to be manually annotated. Thanks to advances in machine learning, a computer is now able to automatically provide accurate timing annotations for a wide variety of audio and lyrics.
AutoLyrixAlign is a system published in 2019 which embody these advancements. Given an audio file and a text file containing associated lyrics, AutoLyrixAlign provides timing information for each word in a complete set of lyrics. Aside from this core feature, AutoLyrixAlign is feature sparse. It is executed via Command-Line and only
LyricManager provides a more user friendly interface to manage and align lyrics for a large collection of audio files, with such features as:
- Two fully functional interfaces: Graphical and Command-line
- Multiple online sources for fetching lyrics
- Multi-song parsing
- Additional audio alignment clean up
- JSON-based output
Resources
- LyricManager Github – Full source code
- AutoLyrixAlign Github – Offline version
- AutoLyrixAlign – Online version
Quick Facts
- Built from the ground up in Python and Qt6 (PySide6)
- Fully shared code-path between GUI and CLI interfaces
- Multi-platform
- Binaries available for Windows and Linux
Connecting audio and video
Automatically aligning lyrics to audio, while impressive, is not going to interest audiences on its own. A system is needed to create captivating visuals based on the lyric timing information. PlanMixPlay natively supports the JSON output that LyricManager generates.
The PlanMixPlay project page offers further details on the importance of synchronizing visuals to audio. A significant benefit of the emergence of automatic aligning is that any song featuring lyrics can now have impressive custom song-specific visuals generated to accompany it.
Acknowledgements
- shargo for quality assurance and supplying ideas during development.
- Chitralekha Gupta, Emre Yilmaz, Haizhou Li for developing AutoLyrixAlign.