DocsAudrey
Audrey
Speech to text for real calls: send a recording, or stream live audio and get text while people are still talking.
What Audrey does
Audrey turns speech into accurate text. It is built for the audio companies actually have: accented speakers, people talking over each other, noisy phone lines. Names, numbers and account details come through intact, and by default numbers, amounts, times and addresses are written the way people write them. It serves contact centres, sales teams and voice agents.
Audrey serves English today. Responses name the model version, for example audrey-0.7.
Two ways to send audio
Both use the same key and the same balance.
Features
- Written forms: phone numbers, amounts, times and dates as people write them, on by default.
- Turn detection: streams end an utterance as soon as the speaker has finished, using turn-1-mini.
- Context prompt: up to 500 characters of context, such as the names and terms you expect.
- Flush: finalise an utterance the moment your own end-of-speech logic decides.
Pricing
From $0.15 per audio hour (batch) · from $0.25 (streaming). Volume pricing available.
See billing and credits for plans and the sign-up credit.