# Soft Story Audio

> Korean text-to-speech for agents and developers. 191 voices (Korean; 102 of them also speak English, Japanese and Chinese), emotion-matched Korean delivery, async jobs for long scripts. Credits are voice-usage units: the API and MCP charge 320 credits per 1,000 characters (rounded up per request; emotion requests on SSA-2.0 use more), and the web studio charges 1 credit per character.

Site: https://softstoryaudio.com — Studio (web editor), pricing, API keys at /dashboard/keys.

## MCP (Model Context Protocol)

- Endpoint: `https://softstoryaudio.com/mcp` — Streamable HTTP, stateless, JSON responses.
- Auth: `Authorization: Bearer <api key>` (keys start with `ssa_`; create one at https://softstoryaudio.com/dashboard/keys; choose 30 days, 90 days (default), 1 year or no expiry). Required on every request, including `initialize`. An expired key is rejected with the same 401 as an invalid key.
- Tools: `list_voices` (free), `synthesize` (320 credits per 1,000 characters, rounded up per request), `get_job` (free), `get_credits` (free), `list_models` (free).
- Claude Code: `claude mcp add --transport http soft-story-audio https://softstoryaudio.com/mcp --header "Authorization: Bearer ssa_..."`
- Cursor / Windsurf / VS Code: add an HTTP MCP server with url `https://softstoryaudio.com/mcp` and the Authorization header.
- Claude Desktop: `npx mcp-remote https://softstoryaudio.com/mcp --header "Authorization: Bearer ssa_..."`

## REST API (OpenAPI 3.1)

- Spec: https://softstoryaudio.com/api/v1/openapi (OpenAPI 3.1 JSON) — docs: https://softstoryaudio.com/docs
- `GET /api/v1/voices` · `POST /api/v1/synthesize` (`mode: "sync"` returns audio inline for one utterance up to 300 characters; otherwise 202 + job) · `GET /api/v1/jobs/{id}?wait=10` · `GET /api/v1/jobs/{id}/audio`
- Rate limits: 120 requests/min, 16 concurrent renders. Failed renders are refunded automatically.

## Rules worth knowing

- Emotion is Korean-only and only the labels listed for a voice (e.g. 차분함 · 기쁨 · 슬픔 · 분노) have an effect.
- Models: `SSA-1.3` (default; newer prosody), `SSA-1.2`, and `SSA-2.0` (beta: emotion acting with intensity; emotions 분노·슬픔·사랑).
- Generated voice that is published must be labelled as AI-generated under Korean law (AI Basic Act).
- Voices are catalog ids and display names only; the people behind them are not identified.

## Key pages

- Korean landing (try voices in the browser): https://softstoryaudio.com/ — English: /en · 中文: /zh · 日本語: /ja
- Pricing (web studio plans and credit packs): https://softstoryaudio.com/pricing
- API pricing (monthly developer plans): https://softstoryaudio.com/pricing/api
- API reference: https://softstoryaudio.com/docs

## Web studio plans (KRW, VAT included, monthly)

- Starter ₩7,900 · Pro ₩20,900 · Pro Speaker ₩53,900 · MAX Speaker ₩163,900 (annual billing 20% off). A free plan gives 3,000 credits.
- API plans: API Starter ₩10,890 · API Growth ₩53,900 · API Scale ₩163,900 per month.

## 한국어 요약

Soft Story Audio 는 한국어 음성 합성(TTS) 서비스입니다. 자체 학습한 한국어 보이스로 광고 · 오디오북 · 영상 더빙용 음성을 감정 톤까지 직접 완성할 수 있습니다.

- 보이스 191개(한국어). 그중 102개는 영어 · 일본어 · 중국어도 말합니다. 감정 연기는 한국어만, 각 보이스에 표시된 감정만 적용됩니다.
- 웹 스튜디오: 1크레딧 = 약 한국어 1글자. 원고를 줄 단위로 합성 · 편집하고 MP3 로 내보냅니다(WAV 는 유료 플랜). 무료 플랜은 3,000크레딧.
- 요금(월, VAT 포함): Starter ₩7,900 · Pro ₩20,900 · Pro Speaker ₩53,900 · MAX Speaker ₩163,900. 연 결제 20% 할인.
- 개발자: REST API(OpenAPI 3.1)와 MCP 서버. API 는 1,000자당 320크레딧. 문서 https://softstoryaudio.com/docs
- 공개하는 AI 생성 음성은 인공지능기본법에 따라 AI 생성물임을 표시해야 합니다.