ElevenLabs Says Cue Cuts Audio Timing Error to 179 Milliseconds
Summary
ElevenLabs details Cue, a forced alignment model that places every character of an audio file on a precise timeline. The company says Cue averages 179 milliseconds of timing error on full recordings, versus 283 to more than 6,000 milliseconds for tested open-source tools, and aligns an hour of audio in about half a second. It covers more than 90 languages.
Key Points
- Researcher Danila Kozlov says Cue is smaller than ElevenLabs' previous aligner.
- Cue places 92.7% of words within 0.3 seconds on JamendoLyrics songs, versus 82.0% for the next best tool.
- ElevenLabs retrained Cue for Georgian and other low-resource languages in under 24 hours after a customer report.