Idea: Use lyrics of a website
Opened this issue · 4 comments
I had the idea to create a similar project.
What I thought about was using existing lyrics of a website like genius.com instead of a speech recognition model. There is also an API (https://lyricsgenius.readthedocs.io/en/master/index.html). Alternatetively, you can use the lyrics correct mistakes of the speech recognition model.
Yeah good idea. I think we need an recognition correction with given lyric.
Is there something similar on the web that has also info for syllables/hypen?
I appreciate this idea - schould give better results in linebreaks and wording
chagpt suggests these web sites for syllables/hypen :-)
https://www.nltk.org/
https://opennlp.apache.org/
https://sapui5.hana.ondemand.com/
https://github.com/EtienneAb3d/SRT-Sync
Synchronize SRT timestamps over an existing accurate transcription
Input 1: SRT with good timestamps and bad-quality text
Input 2: good text-only, or SRT with good text and bad timestamps
Output: SRT with good text and good timestamps
If this works as said, this could be used for improve the results of UltraSinger IMO.