Deepgram-compatible API
Parameters
Configure the session in the query string, as with Deepgram. Deepgram parameters that have no effect here are accepted, so plugins can keep their defaults.
Query parameters
| Parameter | Values |
|---|---|
encoding | linear16, the default; mulaw; or alaw. |
sample_rate | Required: 16000 or 24000 for linear16, 8000 for mulaw and alaw. |
language | A tag from the language list. Leave it out, or set multi, for automatic detection. With a client secret, leave it out or repeat the secret’s language. |
utterance_end_ms | The pause that ends a segment, from 400 to 3200. |
interim_results | true, the default, sends interim Results while a segment grows; false sends only finals. |
model | Accepted, and can stay at your plugin’s default. It must not be empty. |
punctuate, smart_format | Accepted. |
endpointing | Accepted, with nothing to set: segments end after utterance_end_ms. |
channels | 1: audio must have one channel. |
diarize, multichannel, keywords | Refused, with an error that names the parameter. |
Audio formats
| Audio | Query parameters |
|---|---|
| PCM16, 24 kHz | encoding=linear16&sample_rate=24000 |
| PCM16, 16 kHz | encoding=linear16&sample_rate=16000 |
| G.711 μ-law, 8 kHz | encoding=mulaw&sample_rate=8000 |
| G.711 A-law, 8 kHz | encoding=alaw&sample_rate=8000 |
Send audio as binary messages, in chunks of any size, at the speed of speech: the server holds up to 640 ms ahead of real time, and audio sent faster ends the session.