Audio and speech models
Make every spoken word in your archive searchable.
What they do
- Transcription Text from speech, in multiple languages
- Language detection Know which language is spoken, without listening
- Speaker identification Who is speaking and when
- Audio classification Music, applause, crowd noise, gunshots and other sounds
Why it matters
Interviews, rushes, podcasts and calls hold huge value that nobody can search. With a transcript and speaker labels attached to each asset, editors find the exact line in seconds.
Works with the rest of the platform
Transcripts feed text understanding for summaries and answers, and sit alongside video and image results so one search covers what is seen and what is said.
FAQ
Which languages are supported?
Multiple languages. We will confirm your languages in a demo.
Does it translate?
Not at present.