Audio Speech (TTS)

Text-to-speech. Synchronous, returns audio bytes, billed per character. Several voice models, 3,000+ pre-made voices.

POST /v1/audio/speech

Audio Speech (TTS)

cURL

curl --request POST \
  --url https://kymaapi.com/v1/audio/speech \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "model": "<string>",
  "input": "<string>",
  "voice_id": "<string>",
  "response_format": "<string>",
  "voice_settings": {},
  "stream": true
}
'

Synchronous endpoint. Send text, get back audio bytes in one call. Pick a model based on quality needs vs latency budget.

curl -X POST https://kymaapi.com/v1/audio/speech \
  -H "Authorization: Bearer $KYMA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "eleven-multilingual-v2",
    "input": "The first move is what sets everything in motion.",
    "voice_id": "JBFqnCBsd6RMkjVDRZzb",
    "response_format": "mp3_44100_128"
  }' \
  --output speech.mp3

Request

application/json body.

model · string · default: "eleven-multilingual-v2"

One of:

• eleven-v3 — most expressive, audio tags + emotional range, 70+ languages ($0.405/1K char)
• eleven-multilingual-v2 — hero quality, 29 languages, expressive ($0.405/1K char)
• eleven-flash-v2-5 — ~75ms time-to-first-byte, real-time agents ($0.20/1K char)
• eleven-turbo-v2-5 — balanced quality + speed ($0.20/1K char)

input · string · required

Text to synthesize. Max 5000 characters per request — chunk longer text client-side. Also accepts the alias text for OpenAI compatibility.

voice_id · string · required

ElevenLabs voice id — opaque string from GET /v1/audio/voices. Also accepts the alias voice. There is no global default; pick one explicitly.

response_format · string · default: "mp3_44100_128"

Audio format. mp3_44100_128 (default), mp3_44100_192, mp3_22050_32, pcm_16000/22050/24000/44100, ulaw_8000.

voice_settings · object

Optional fine-tuning: stability (0–1), similarity_boost (0–1), style (0–1), use_speaker_boost (boolean).

stream · boolean · default: "false"

Opt-in low-latency mode. When true, audio is delivered progressively as it is synthesized — time-to-first-audio drops to ~0.4s (vs ~1.8s for the default full-buffer path). Transparent to your code: the response is still a single complete audio/mpeg stream you can pipe straight to disk or a player. Currently applies to the MiniMax speech models; ignored by models that don’t support progressive synthesis.

Response

200 OK with raw audio bytes. Billing rides on X-Kyma-* response headers — the body stays a clean audio file you can pipe straight to disk or play.

Header	What
Content-Type	matches the requested format (audio/mpeg for mp3, etc.)
X-Kyma-Model	resolved model id
X-Kyma-Chars-Billed	input char count used for pricing
X-Kyma-Cost-USD	actual cost charged
X-Kyma-Balance-USD	remaining balance

Errors

Status	error.code	When
400	not_a_tts_model	model is not a TTS SKU
400	voice_required	voice / voice_id missing
400	input_too_long	input text > 5000 chars
400	invalid_request	invalid JSON or missing input
401	auth_error	missing or invalid API key
402	billing_error	balance too low
502	provider_error	upstream TTS provider failure

See also

• GET /v1/audio/voices — browse the voice library
• POST /v1/audio/music — generative music
• POST /v1/audio/sfx — sound effects