Vocal splitting

If Spleeter is installed, Lyric Timer can use it to run speech recognition on the vocals only. This might improve quality.

To use this feature, install Spleeter and add the -s/--split-vocals flag: lyrictimer -s ...

More formats

In addition to .srt and .vtt, Lyric Timer can now export in .lrc. The format is now selected with the -f/--format flag: lyrictimer -f lrc ...

This means that the --vtt flag is deprecated and will be removed in the next version. Please replace it with -f vtt.

If word highlighting is on, the output file will be in enhanced/A2 LRC format with word-level timestamps. Keep in mind that not all media players support this format.

Raw exports

This is an export format that might help developers or advanced users, and is especially useful if the program fails to recognize something. It uses a special .ltr (Lyric Timer raw) extension, and can be selected like any other format (lyrictimer -f ltr).

Raw files are JSON files containing:

the version and configuration (if any advanced features, like phoneme-based correction, are enabled),
the real lyrics of the song,
the transcription output.

Eventually it will be possible to use these raw files to process lyrics again and export them to a different format without having to re-run speech recognition.

Minor changes

If text recognition returns an empty sentence, it's now skipped to avoid errors with phoneme-based correction and make the file smaller. ( d9b0045e).