Fish Audio S2.1 Pro API

Read this first: billed per byte, not per character

This model is billed on the UTF-8 byte length of your input text, while every other speech model here bills per character. A Latin letter is one byte, a Chinese character is three, an emoji is four. A thousand Chinese characters therefore costs three times a thousand English ones. If you are switching from another speech model, re-estimate CJK workloads on bytes rather than carrying the character count across.

Charged only on success. Your first 20 voice clones are free.

Choosing a tier

Code examples

# ─── 合成一段语音 ───────────────────────────────────────
curl -X POST https://api.apimodels.app/v1/audio/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "fish-s2.1-pro",
    "text": "Hello from apimodels.",
    "format": "mp3"
  }'
# { "code": 200, "data": { "taskId": "clxxx", "state": "pending" } }


# ─── 用克隆出来的音色说话 ────────────────────────────────
curl -X POST https://api.apimodels.app/v1/audio/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "fish-s2.1-pro",
    "text": "Now I speak in the cloned voice.",
    "reference_id": "YOUR_VOICE_ID",
    "format": "wav",
    "prosody": { "speed": 1.2 }
  }'


# ─── 取结果 ─────────────────────────────────────────────
curl "https://api.apimodels.app/v1/audio/generations?task_id=TASK_ID" \
  -H "Authorization: Bearer YOUR_API_KEY"
# { "code": 200, "data": { "state": "completed", "resultUrls": ["https://r2.apimodels.app/..."] } }

Request parameters

Picking a voice

Without reference_id you get Fish's default voice, which can sound flat outside English. Pick one from the Fish voice library and pass its voice_id as reference_id:

GET https://api.apimodels.app/v1/audio/voices/fish/library?source=licensed&language=en&q=calm&page=1
Authorization: Bearer YOUR_API_KEY

Each voice comes back with voice_id, name, languages, preview_url, licensed and uses. Licensed voices are fewer (some languages, such as Arabic, have none yet). Community voices are uploaded by Fish Audio users and some imitate real people, so confirm you have the rights before commercial use. You can also clone your own voice.

No API key at hand? GET https://api.apimodels.app/v1/voices/catalog?provider=fish&source=licensed&language=ja returns the same ids, preview URLs and languages from a daily-refreshed catalog (parameters in the Voices API docs), or browse and preview them in the voice library.

Pricing for Fish licensed voices

Fish Audio licensed voices (licensed: true in the voice library, or anything listed with source=licensed) are billed at twice the rate and are not available on the evaluation tier: sending one to fish-s2.1-pro-free returns a 400. Community voices and your own clones are billed at the normal rate.

Notes

• Your first 20 voice clones are free; past 20 stored voices each extra slot is $1 and deleting one frees it again. Send a clean sample of ten seconds or more and a voice id comes back in roughly 5 to 20 seconds.
• All 83 languages share one model id — no per-language endpoint and no premium for non-English speech.
• Async: create a task for a taskId, then poll GET ?task_id= for the result.
• Result files are deleted after 30 days; download or re-host them in time.
• Charged only on success.

Try it in the Playground