Start speech recognition
POST /server/v1/asr/start
Authentication: required (see Overview)
Start real-time speech recognition for a channel to transcribe what’s said into text, for meeting minutes, captions, and search.
- Calling it again for the same channel doesn’t start a second recognition session
- It recognizes audio from every user in the channel whose mic is on; results carry uid and name to identify the speaker
- Query transcription results by channel with “List speech recognition sentences”
string
required
Channel name
Example:
firedata is null
Response example:
Stop speech recognition
POST /server/v1/asr/stop
Authentication: required (see Overview)
Stop speech recognition for a channel. Transcription results already produced are not deleted and can still be queried with “List speech recognition sentences”.
Stops automatically when the channel is destroyed, so you don’t need to stop it manually first.
Request parameters
string
required
Channel name
Example:
firedata is null
Response example:
List speech recognition sentences
POST /server/v1/asr/list-sentence
Authentication: required (see Overview)
List a channel’s transcription results with pagination; each item is one sentence, with the speaker and time.
- Read the results in created_at order to reconstruct the full conversation; this is the data source for generating meeting minutes
- Results are kept after the channel is destroyed and can be queried afterward
string
required
Channel name
Example:
firestring
Sort order
array<string>
General search; can search transcription content
integer
Page number, starting from 1
Example:
1integer
Page size
Example:
10string
Channel name
Example:
firestring
Speaker’s user ID
Example:
1001string
Speaker’s display name
Example:
Alicestring
Transcribed text of one sentence
Example:
Let's finalize this plan next weekinteger
Time the sentence was produced, Unix timestamp in seconds
Example:
1718250918