Skip to main content

​
A5S Async v1

a5s-async-v1 · REST · batch guide
  • Recorded audio or video in any common format, from 0.1 s to 10 hours.
  • Speaker labels (speaker_0, speaker_1, …), word timestamps, utterances, SRT/VTT subtitles.
  • English only: language_code accepts en and regional variants (en-US, en-GB, …); any other value returns 400 unsupported_language.
  • Text is verbatim and lowercase without punctuation; disfluencies (“uh”, “um”) are kept.
  • Typical turnaround: well under a minute for most files once processing starts.

​
A5S v2 Streaming

a5s-v2-streaming · WebSocket · streaming guide
  • Live 16 kHz mono PCM16 audio in, partial and final text out as people speak.
  • Punctuated, cased text. No speaker labels.
  • Optional custom vocabulary to bias recognition toward names and terms.
  • English only: the start message’s language accepts en and regional variants (en-US, en-GB, …); any other value is refused with unsupported_language.
  • Sessions last up to 5 minutes.

​
Choosing

​
Languages

Both models support English today. Multilingual support is planned and will be available soon. Accepted codes: en, en-US, en-GB, en-AU, en-CA, en-IN, en-IE, en-NZ, en-ZA (case-insensitive; _ works too).

​
Access

Accounts can use both models by default. An account may be limited to one model. Requests to a model your account can’t use return 403 model_access_denied (streaming: an error message with that code, then close code 4403). GET /v1/account shows enabled for each model. To get access, talk to sales. GET /v1/models lists the models.