Voice Design

Generate candidate voices from a prompt

POST /v1/voice-design

Voice Design

cURL

curl --request POST \
  --url https://api.fish.audio/v1/voice-design \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --header 'model: voice-design-1' \
  --data '{
    "instruction": "Warm, confident studio narrator with a natural tone",
    "reference_text": "Welcome to Fish Audio.",
    "language": "en",
    "n": 2,
    "speed": 1,
    "num_step": 32,
    "guidance_scale": 2,
    "instruct_guidance_scale": 0,
    "seed": 42
  }'

import requests

url = "https://api.fish.audio/v1/voice-design"

payload = {
    "instruction": "<string>",
    "reference_text": "<string>",
    "language": "<string>",
    "n": 2,
    "speed": 1,
    "num_step": 32,
    "guidance_scale": 2,
    "instruct_guidance_scale": 0,
    "seed": 123
}
headers = {
    "model": "<model>",
    "Authorization": "Bearer <token>",
    "Content-Type": "application/json"
}

response = requests.post(url, json=payload, headers=headers)

print(response.text)

const options = {
  method: 'POST',
  headers: {
    model: '<model>',
    Authorization: 'Bearer <token>',
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    instruction: '<string>',
    reference_text: '<string>',
    language: '<string>',
    n: 2,
    speed: 1,
    num_step: 32,
    guidance_scale: 2,
    instruct_guidance_scale: 0,
    seed: 123
  })
};

fetch('https://api.fish.audio/v1/voice-design', options)
  .then(res => res.json())
  .then(res => console.log(res))
  .catch(err => console.error(err));

<?php

$curl = curl_init();

curl_setopt_array($curl, [
  CURLOPT_URL => "https://api.fish.audio/v1/voice-design",
  CURLOPT_RETURNTRANSFER => true,
  CURLOPT_ENCODING => "",
  CURLOPT_MAXREDIRS => 10,
  CURLOPT_TIMEOUT => 30,
  CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
  CURLOPT_CUSTOMREQUEST => "POST",
  CURLOPT_POSTFIELDS => json_encode([
    'instruction' => '<string>',
    'reference_text' => '<string>',
    'language' => '<string>',
    'n' => 2,
    'speed' => 1,
    'num_step' => 32,
    'guidance_scale' => 2,
    'instruct_guidance_scale' => 0,
    'seed' => 123
  ]),
  CURLOPT_HTTPHEADER => [
    "Authorization: Bearer <token>",
    "Content-Type: application/json",
    "model: <model>"
  ],
]);

$response = curl_exec($curl);
$err = curl_error($curl);

curl_close($curl);

if ($err) {
  echo "cURL Error #:" . $err;
} else {
  echo $response;
}

package main

import (
	"fmt"
	"strings"
	"net/http"
	"io"
)

func main() {

	url := "https://api.fish.audio/v1/voice-design"

	payload := strings.NewReader("{\n  \"instruction\": \"<string>\",\n  \"reference_text\": \"<string>\",\n  \"language\": \"<string>\",\n  \"n\": 2,\n  \"speed\": 1,\n  \"num_step\": 32,\n  \"guidance_scale\": 2,\n  \"instruct_guidance_scale\": 0,\n  \"seed\": 123\n}")

	req, _ := http.NewRequest("POST", url, payload)

	req.Header.Add("model", "<model>")
	req.Header.Add("Authorization", "Bearer <token>")
	req.Header.Add("Content-Type", "application/json")

	res, _ := http.DefaultClient.Do(req)

	defer res.Body.Close()
	body, _ := io.ReadAll(res.Body)

	fmt.Println(string(body))

}

HttpResponse<String> response = Unirest.post("https://api.fish.audio/v1/voice-design")
  .header("model", "<model>")
  .header("Authorization", "Bearer <token>")
  .header("Content-Type", "application/json")
  .body("{\n  \"instruction\": \"<string>\",\n  \"reference_text\": \"<string>\",\n  \"language\": \"<string>\",\n  \"n\": 2,\n  \"speed\": 1,\n  \"num_step\": 32,\n  \"guidance_scale\": 2,\n  \"instruct_guidance_scale\": 0,\n  \"seed\": 123\n}")
  .asString();

require 'uri'
require 'net/http'

url = URI("https://api.fish.audio/v1/voice-design")

http = Net::HTTP.new(url.host, url.port)
http.use_ssl = true

request = Net::HTTP::Post.new(url)
request["model"] = '<model>'
request["Authorization"] = 'Bearer <token>'
request["Content-Type"] = 'application/json'
request.body = "{\n  \"instruction\": \"<string>\",\n  \"reference_text\": \"<string>\",\n  \"language\": \"<string>\",\n  \"n\": 2,\n  \"speed\": 1,\n  \"num_step\": 32,\n  \"guidance_scale\": 2,\n  \"instruct_guidance_scale\": 0,\n  \"seed\": 123\n}"

response = http.request(request)
puts response.read_body

200

{
  "candidates": [
    {
      "id": "<string>",
      "index": 1,
      "audio_base64": "<string>",
      "sample_rate": 123,
      "duration_ms": 1,
      "text": "<string>",
      "instruct": "<string>",
      "language": "<string>"
    }
  ]
}

{
  "status": 123,
  "message": "<string>"
}

{
  "status": 123,
  "message": "<string>"
}

[
  {
    "loc": [
      "<string>"
    ],
    "type": "<string>",
    "msg": "<string>",
    "ctx": "<string>"
  }
]

This endpoint only accepts application/json.

You must include the model: voice-design-1 header. Extra request fields are rejected.

A successful request returns generated voice candidates with audio_base64 audio payloads. Decode the base64 value to write the candidate audio to a file.

Example

curl --request POST https://api.fish.audio/v1/voice-design \
  --header "Authorization: Bearer $FISH_API_KEY" \
  --header "Content-Type: application/json" \
  --header "model: voice-design-1" \
  --data '{
    "instruction": "Warm, confident studio narrator with a natural tone",
    "reference_text": "Welcome to Fish Audio.",
    "language": "en",
    "n": 2,
    "speed": 1,
    "num_step": 32,
    "guidance_scale": 2,
    "instruct_guidance_scale": 0,
    "seed": 42
  }'

Usage notes

• instruction is required and must be 1 to 2000 characters.
• reference_text is optional preview text and can be up to 150 characters.
• n controls how many candidates are returned. The supported range is 1 to 4.
• seed is optional and can help reproduce candidate generation.
• The endpoint is stateless: it does not create batches, samples, voice models, or presigned URLs.
• Billing happens once per successful generation request, not once per candidate.

Authorizations

Authorization · string · header · required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Headers

model · string · default: voice-design-1 · required

Specify which voice-design model to use.

Allowed value: "voice-design-1"

Body

application/json

Request body for synchronous voice design generation. The endpoint returns generated voice candidates with base64-encoded audio.

instruction · string · required

Voice design prompt. Must contain 1 to 2000 characters.

Required string length: 1 - 2000

reference_text · string | null

Optional text used as reference content for the generated voice.

Maximum string length: 150

language · string | null

Optional BCP-47 language hint, such as en, zh, or ja.

n · integer · default: 2

Number of voice candidates to generate.

Required range: 1 <= x <= 4

speed · number · default: 1

Speaking speed multiplier for candidate generation.

Required range: x <= 3

num_step · integer · default: 32

Number of diffusion steps used by the voice-design model.

Required range: 1 <= x <= 128

guidance_scale · number · default: 2

Classifier-free guidance scale. Higher values follow the prompt more strongly.

Required range: x >= 0

instruct_guidance_scale · number · default: 0

Instruction guidance scale for prompt conditioning.

Required range: x >= 0

seed · integer | null

Optional deterministic seed for candidate generation.

Response

200

application/json

Request fulfilled, document follows

candidates · VoiceDesignCandidate · object[] · required

Generated voice candidates.

Show child attributes