Skip to main content
POST
Create speech

Authorizations

Authorization
string
header
required

Your API key

Body

application/json
input
string
required

The text. Long texts are split into chunks and streamed (up to 20,000 characters).

Example:

"Siparişiniz 28 Ekim günü saat 14.30da teslim edilecek."

model
string
default:mihu-ses-v1

mihu-ses-v1 (native), or an OpenAI name: tts-1, tts-1-hd, gpt-4o-mini-tts.

language
string

tr, en, de, fr, es, it, ru, or auto (detect from the text). Without it: from the voice, else Turkish (native) or detected (OpenAI).

Example:

"tr"

voice
string

Voice id or name, an OpenAI voice name, or a custom voice id. See Voices.

Example:

"alp"

response_format
enum<string>

pcm is raw 16-bit little-endian 24 kHz mono. Default pcm (native), mp3 (OpenAI).

Available options:
pcm,
wav,
mp3,
opus,
aac,
flac
Example:

"wav"

speed
number
default:1

0.25 to 4.0: tempo, pitch unchanged.

stream_format
enum<string>
default:audio

audio: the audio bytes. sse: Server-Sent Events.

Available options:
audio,
sse
normalize
boolean
default:true

Numbers, amounts, dates, times and abbreviations written out before synthesis.

verify
boolean
default:false

Turkish: a recognizer checks each chunk; first audio comes about 2 s later. Not for live calls.

instructions
string

Accepted and ignored (OpenAI style prompt).

Response

The audio, streamed. Headers: X-Sample-Rate, X-Language, X-Voice, X-Model.

The response is of type file.