Sync Model

Youka offers different methods to match the lyrics to the music in your karaoke tracks, known as sync models. There are two main types: transcription and alignment.

Model Types #

• Transcription models listen to the song and try to write down the lyrics like a person would, using technology similar to voice recognition. However, the words they write down might not be perfectly accurate.

• Alignment models need you to provide the lyrics. They then try to match these lyrics with the song. If the lyrics you provide don't match the song exactly (like if a chorus is missing), the timing might be a bit off in parts.

Available Models #

AudioShakeAI (Transcription) #

This is the top choice and works by first writing down the lyrics and then matching them to the music. It's only available if you're a Pay-Per-Use user or using a trial, and it might take up to 10 minutes to work.

AudioShakeAI (Alignment) #

This comes next and matches the lyrics you provide to the music. If there are mistakes in the lyrics, the timing might not be perfect. It also takes up to 5 minutes.

Wav2Vec2 (Alignment) #

This is the third option and works like AudioShakeAI but supports almost all languages. It's faster, usually finishing in up to 30 seconds.

Whisper (Transcription) #

This is the fourth option and also listens to the music to write down the lyrics before syncing them. The lyrics might not be spot-on, but the timing should be decent, taking up to 2 minutes to complete.