Skip to main content
POST
Unmute STT Bridge

Authorizations

Authorization
string
header
required

API key issued by SLNG. Pass as Authorization: Bearer <token>.

Path Parameters

model_variant
enum<string>
required

Target STT model. Must support WebSocket streaming with final_transcript.

Available options:
deepgram/nova:3,
deepgram/nova:3-medical,
gradium/stt:default,
reson8/reson8stt:v1,
sarvam/saaras:v3,
slng/deepgram/nova:3-en,
slng/deepgram/nova:3-hi,
slng/deepgram/nova:3-id,
slng/deepgram/nova:3-kn,
slng/deepgram/nova:3-mr,
slng/deepgram/nova:3-te,
slng/speechmatics/realtime:v2,
soniox/speech-ai:rt-v5

Body

Unmute STT request. Provide either audio (multipart upload) or url (publicly accessible HTTPS URL).

audio
file

Audio file (multipart) or base64-encoded audio (JSON).

url
string

Publicly accessible audio URL.

Example:

"https://docs.slng.ai/audio/hello.wav"

language
string

Language code for recognition.

sample_rate
enum<integer>

Audio sample rate in Hz.

Available options:
8000,
16000,
22050,
24000,
32000,
48000
encoding
enum<string>

Audio encoding format.

Available options:
linear16,
mp3,
opus
enable_partials
boolean

Enable partial transcription results.

Response

Transcription successful. The body is the target model's native HTTP transcription response, returned unchanged.

The target model's native HTTP transcription response, passed through unchanged. For example, Deepgram Nova returns results.channels[].alternatives[]. See the direct endpoint of the selected model for its exact shape.