Synthesize speech

POST

speak

Handler for the /speak endpoint

curl --request POST \
  --url https://api.sayna.ai/speak \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "text": "Hello, world!",
  "tts_config": {
    "model": "aura-asteria-en",
    "provider": "deepgram",
    "audio_format": "linear16",
    "connection_timeout": 30,
    "pronunciations": [
      {
        "pronunciation": "A P I",
        "word": "API"
      }
    ],
    "request_timeout": 60,
    "sample_rate": 24000,
    "speaking_rate": 1,
    "voice_id": "aura-asteria-en"
  }
}
'

Kick off one-shot synthesis jobs for short responses. This endpoint reuses the same provider layer and cache that powers the streaming WebSocket experience.

Reuse identical tts_config inputs to hit the cache and avoid extra provider round-trips.

Authorizations

Authorization

string

header

required

JWT token obtained from the authentication service. Required when AUTH_REQUIRED is enabled.

Body

application/json

Request body for the speak endpoint

text

string

required

The text to synthesize

Example:

"Hello, world!"

tts_config

object

required

TTS configuration (without API key)

Show child attributes

tts_config.model

string

required

Model to use for TTS

Example:

"aura-asteria-en"

tts_config.provider

string

required

Provider name (e.g., "deepgram")

Example:

"deepgram"

tts_config.audio_format

string | null

Audio format preference

Example:

"linear16"

tts_config.connection_timeout

integer<int64> | null

Connection timeout in seconds

Required range: x >= 0

Example:

30

tts_config.pronunciations

object[]

Pronunciation replacements to apply before TTS

Show child attributes

tts_config.pronunciations.pronunciation

string

required

Pronunciation to use instead

Example:

"A P I"

tts_config.pronunciations.word

string

required

Word to replace

Example:

"API"

tts_config.request_timeout

integer<int64> | null

Request timeout in seconds

Required range: x >= 0

Example:

60

tts_config.sample_rate

integer<int32> | null

Sample rate preference

Required range: x >= 0

Example:

24000

tts_config.speaking_rate

number<float> | null

Speaking rate (0.25 to 4.0, 1.0 is normal)

Example:

1

tts_config.voice_id

string | null

Voice ID or name to use for synthesis

Example:

"aura-asteria-en"

Response

Audio generated successfully

LiveKit tokenGenerates a LiveKit JWT token for a participant to join a specific room. # Arguments * `state` - Shared application state containing LiveKit configuration * `request` - Token request with room name and participant details # Returns * `Response` - JSON response with token or error status # Errors * 400 Bad Request - Invalid request data (empty fields) * 500 Internal Server Error - LiveKit service not configured or token generation failed

⌘I

Handler for the /speak endpoint

curl --request POST \
  --url https://api.sayna.ai/speak \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "text": "Hello, world!",
  "tts_config": {
    "model": "aura-asteria-en",
    "provider": "deepgram",
    "audio_format": "linear16",
    "connection_timeout": 30,
    "pronunciations": [
      {
        "pronunciation": "A P I",
        "word": "API"
      }
    ],
    "request_timeout": 60,
    "sample_rate": 24000,
    "speaking_rate": 1,
    "voice_id": "aura-asteria-en"
  }
}
'

Overview

REST endpoints

Synthesize speech

Authorizations

Body

Response