Desert Ant Labs's banner

Desert Ant Labs

@desertantlabs • 4,224 subscribers

European AI lab building opinionated on-device models and SDKs. Little brains in every product.

Shorts

WhisperX uses a 378MB model to time English words. Align does it in 0.7MB. Give it any transcript and it corrects each word’s start and end time against the audio, so captions highlight every word exactly as it’s spoken. On 258 word boundaries checked by hand, Align was off by 45ms on average. WhisperX was off by 53.5ms.

WhisperX uses a 378MB model to time English words. Align does it in 0.7MB. Give it any transcript and it corrects each word’s start and end time against the audio, so captions highlight every word exactly as it’s spoken. On 258 word boundaries checked by hand, Align was off by 45ms on average. WhisperX was off by 53.5ms.

22,687 views

Videos

Anya Rossi

sweetdream.ai

SweetDream.ai•Sponsored•Livecam

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Exclusive private shows

1.2k viewers online

Private Show

Join now for exclusive access

Free preview available • Premium content

You asked for speaker diarization, so we built it. Who split this 43-minute podcast by speaker in 4.9s on an M5 chip, including the moments two people talk at the same time. The recording never left the machine. Who is in our Swift, Kotlin and JavaScript SDKs, free up to 100k monthly active devices:

You asked for speaker diarization, so we built it. Who split this 43-minute podcast by speaker in 4.9s on an M5 chip, including the moments two people talk at the same time. The recording never left the machine. Who is in our Swift, Kotlin and JavaScript SDKs, free up to 100k monthly active devices:

Desert Ant Labs

47,217 views • 24 days ago

Tongue detects the language from three words, with a 2MB model on the device. On three-word samples from FLORES-200, Tongue scores 0.933. A 293MB language detector scores 0.887. Tongue covers 84 languages across 31 scripts.

Tongue detects the language from three words, with a 2MB model on the device. On three-word samples from FLORES-200, Tongue scores 0.933. A 293MB language detector scores 0.887. Tongue covers 84 languages across 31 scripts.

Desert Ant Labs

11,965 views • 1 month ago

No more content to load