STT Control
These APIs pause and resume real-time Speech-to-Text (STT) inference for a specific stream at runtime, without restarting the server or recreating the stream.
For full STT configuration, see Realtime Speech-to-Text.
Enable STT
Resumes STT inference for the stream. Audio frames are passed to the Whisper model and subtitle cues are generated again.
Request
POST v1/vhosts/{vhost}/apps/{app}/streams/{stream}:enableStt
Responses
200 Ok
401 Unauthorized
404 Not Found
The given vhost, app, or stream name could not be found, or no STT encoder exists for this stream.
Body
{
"statusCode": 404,
"message": "Could not find STT encoder for stream: [default/app/stream]"
}
503 Service Unavailable
The Transcoder module is not running.
Disable STT
Pauses STT inference. Audio frames are dropped without processing, and subtitle renditions remain in the stream but stop receiving new cues. The underlying STT model and GPU allocations remain loaded.
Request
POST v1/vhosts/{vhost}/apps/{app}/streams/{stream}:disableStt
Responses
200 Ok
401 Unauthorized
404 Not Found
The given vhost, app, or stream name could not be found, or no STT encoder exists for this stream.
Body
{
"statusCode": 404,
"message": "Could not find STT encoder for stream: [default/app/stream]"
}
503 Service Unavailable
The Transcoder module is not running.
Get STT Status
Returns the current enabled state and configuration of all active STT renditions for the stream.
Request
POST v1/vhosts/{vhost}/apps/{app}/streams/{stream}:sttStatus
Responses
200 Ok
The request has succeeded
Header
Content-Type: application/json
Body
{
"statusCode": 200,
"message": "OK",
"response": {
"enabled": true,
"renditions": [
{
"codec": "WHISPER",
"label": "Korean",
"model": "whisper_model/ggml-small.bin",
"language": "auto",
"translation": "false"
},
{
"codec": "WHISPER",
"label": "English",
"model": "whisper_model/ggml-small.bin",
"language": "auto",
"translation": "true"
}
]
}
}
# statusCode
Same as HTTP Status Code
# message
A human-readable description of the response code
# response.enabled
true if STT inference is currently running, false if paused
# response.renditions
List of active STT encoder instances for this stream
## codec
STT engine in use. Currently always "whisper".
## label
Subtitle rendition label this instance writes to.
## model
Model file path used by this instance.
## language
Configured source language ("auto" or a language code).
## translation
"true" if translation to English is enabled.
401 Unauthorized
503 Service Unavailable
The Transcoder module is not running.