Audio

POST/v1/audio/speech

Synthesize speech and transcribe audio through unified endpoints.

Speech synthesis may return binary audio or a JSON asynchronous task. Transcription uses multipart/form-data. Provide exactly one of input or prompt for speech synthesis.

Read the complete audio guide for sync/async TTS, file saving, transcription, and errors

Speech synthesis fields

Field	Type	Required	Description
`model`	string	Yes	A speech synthesis model ID returned by /v1/models.
`input`	string	No	Text to convert into speech; use it instead of prompt.
`prompt`	string	No	Prompt for audio or music models; use it instead of input.
`voice`	string	No	A voice ID supported by the model; music models usually omit it.
`response_format`	string	No	Requested audio format; accepted values depend on the model.
`instructions`	string	No	Delivery or tone instructions when supported by the model.
`speed`	number	No	Speech speed for compatible TTS models.
`audio_references`	string[]	No	Seed Audio reference audio URLs; up to three.
`image_urls`	string[]	No	Seed Audio reference image URL; mutually exclusive with audio_references.
`format`	string	No	Seed Audio output format, such as wav or mp3.
`sample_rate`	integer	No	Seed Audio sample rate, such as 24000 or 48000.
`speech_rate`	number	No	Seed Audio speech rate.
`loudness_rate`	number	No	Seed Audio loudness adjustment.
`pitch_rate`	number	No	Seed Audio pitch adjustment.
`custom_mode`	boolean	No	Suno custom-mode switch.
`instrumental`	boolean	No	Whether Suno should generate instrumental audio.
`style`	string	No	Suno style description; required when custom_mode=true.
`title`	string	No	Suno track title; required when custom_mode=true.
`duration`	integer	No	Forwarded only for Suno v5.5/v6 with custom_mode=true; acceptance depends on the model.
`vocal_gender`	string	No	Suno vocal-gender preference.
`negative_tags`	string	No	Suno tags to avoid.
`style_weight`	number	No	Suno style weight.
`weirdness_constraint`	number	No	Suno weirdness constraint.
`audio_weight`	number	No	Suno reference-audio weight.
`persona_id`	string	No	Suno Persona ID.
`persona_model`	string	No	Suno Persona model.

Speech synthesis example

curl https://api.noxgate.net/v1/audio/speech \
  -H "Authorization: Bearer sk-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{"model":"suno-v5.5","prompt":"Bright adventure game music","custom_mode":false,"instrumental":true}'

curl https://api.noxgate.net/v1/audio/speech \
  -H "Authorization: Bearer sk-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{"model":"doubao-seed-audio-1-0","prompt":"Gentle rain falling in a forest"}'

curl https://api.noxgate.net/v1/audio/speech \
  -H "Authorization: Bearer sk-your-api-key" \
  -H "Content-Type: application/json" \
  -d '{"model":"YOUR_MODEL_ID","input":"Hello from NoxGate","voice":"YOUR_VOICE_ID"}' \
  -o speech.mp3

Speech synthesis response

Inspect Content-Type first. Save audio responses as bytes; parse application/json as a task, retain its id, and poll the retrieval endpoint.

Retrieve an asynchronous speech task

GET/v1/audio/speech/:task_id

Poll with the id from the JSON creation response. succeeded, failed, and canceled are terminal states; clients must enforce a maximum wait time.

curl "https://api.noxgate.net/v1/audio/speech/audio_task_example" \
  -H "Authorization: Bearer sk-your-api-key"

Audio transcription

POST/v1/audio/transcriptions

Upload an audio file and receive its transcription.

Transcription fields

Field	Type	Required	Description
`model`	string	Yes	A transcription model ID returned by /v1/models.
`file`	file	Yes	The audio file uploaded as multipart data.
`Other form fields`	string	No	Other text form fields are passed to the Provider; use only fields supported by the selected model.

Transcription examples

curl https://api.noxgate.net/v1/audio/transcriptions \
  -H "Authorization: Bearer sk-your-api-key" \
  -F "model=YOUR_MODEL_ID" \
  -F "file=@./audio.mp3"

import requests

with open("audio.mp3", "rb") as audio:
    response = requests.post(
        "https://api.noxgate.net/v1/audio/transcriptions",
        headers={"Authorization": "Bearer sk-your-api-key"},
        data={"model": "YOUR_MODEL_ID"},
        files={"file": ("audio.mp3", audio, "audio/mpeg")},
    )
print(response.json())

import { readFile } from "node:fs/promises";

const form = new FormData();
form.append("model", "YOUR_MODEL_ID");
form.append("file", new Blob([await readFile("audio.mp3")], { type: "audio/mpeg" }), "audio.mp3");

const response = await fetch("https://api.noxgate.net/v1/audio/transcriptions", {
  method: "POST",
  headers: { Authorization: "Bearer sk-your-api-key" },
  body: form,
});
console.log(await response.json());

Transcription response

The gateway preserves the Provider transcription payload. JSON routes commonly return a text field, but the exact shape depends on the selected model.

{"text":"Hello from NoxGate"}

Usage notes

This table combines fields from multiple model families; do not send every field. Suno and Seed Audio use prompt without voice. Traditional TTS uses input and a voice when required by the model.
Voice IDs, formats, file types, and optional fields are constrained by the selected model.
The speech example uses -o to save binary output; do not parse it as JSON.
Transcription requires multipart/form-data and cannot receive a local file path inside JSON.