YOSUGA Developer APIs

Twelve caption presets, every plan

Six of them, cropped to the caption area of the 1080p sample. Portrait and landscape samples of all twelve are linked from GET /v1/caption-presets.

Quick start

Create a job, poll it, download the files. Keys and billing are handled by RapidAPI; the request below is exactly what its test console sends.

1. Create a job

curl -X POST https://video-subtitle-transcript-api1.p.rapidapi.com/v1/jobs \
  -H "X-RapidAPI-Key: $RAPIDAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "source_url": "https://caption.yosuga.dev/samples/video/sample-30s.mp4",
    "language": "ja",
    "mode": "clean",
    "outputs": ["json", "srt", "vtt", "ass"],
    "caption_style": { "preset": "clean_outline" },
    "cut_candidates": true
  }'

# 202 {"id": "job_…", "status": "queued", "poll_url": "/v1/jobs/job_…"}

2. Poll until completed

curl https://video-subtitle-transcript-api1.p.rapidapi.com/v1/jobs/job_… \
  -H "X-RapidAPI-Key: $RAPIDAPI_KEY"

{
  "status": "completed",
  "meta": { "duration_sec": 30.7, "language_detected": "ja", "preset": "clean_outline" },
  "transcript": { "segments": [ { "start": 0.3, "end": 1.0, "text": "こんにちは。", "words": [ … ] } ] },
  "captions": { "cues": [ { "index": 1, "start": 0.3, "end": 1.0, "text": "こんにちは。" } ] },
  "cut_candidates": [ { "type": "filler", "start": 5.7, "end": 6.1, "text": "えー" } ],
  "files": { "srt": "https://…", "vtt": "https://…", "ass": "https://…", "transcript": "https://…" },
  "billing": { "minutes": 1 }
}

Files up to 20 MB can be posted directly to /v1/jobs/upload; larger files (up to 2 GB) go through a signed upload URL from /v1/uploads. Set callback_url on PRO and above to receive a signed webhook instead of polling. Full reference: OpenAPI 3.1.

How billing works

Minutes, rounded upA 10:30 video is 11 minutes. Requests for polling are separate and generous (10,000 a month on the free tier).

Charged once, on completionMinutes are counted the first time a completed result is returned. Polling while a job runs costs nothing.

Failures are freeBad URL, no audio, over the plan's duration limit or an engine error: you get status: failed with a reason, and no Minutes.

APIs

Available on RapidAPI

Video Subtitle & Transcript API

Transcribe video or audio and get SRT, VTT and ASS subtitles with Japanese and English line-breaking, a filler/silence cut list, and burned-in captions.

Word-level timestamps; clean or verbatim transcripts
12 caption presets, previews, burn-in with plan-based watermarking
Async jobs with polling or signed webhooks; files up to 2 GB
Japanese and English in production, 85+ languages in beta

View on RapidAPI OpenAPI 3.1 Status

In development

Japan Business Data API

Furigana and romaji for Japanese names, postal code and address normalization, and corporate number lookup, designed for forms and CRM data in and about Japan.

Planned for release after the caption API. Details will be published here.

In development

Ad Creative Intelligence API

Send an ad (text, image or video URL) or a landing page and get a structured breakdown of why it works: hook type, awareness stage, offer structure, proof, risk reversal and CTA, scored against a direct-response rubric, with rewrite and script variants for your own product.

Analysis and copy only, no media generation. Planned for release after the Japan Business Data API.

Questions people ask first

How long does a job take?: Usually well under a tenth of the media duration: about a minute for a 10-minute video, about 7 minutes for a 2-hour recording. Burn-in adds roughly a fifth of the video length. If the speech engine is busy, the job waits and retries instead of failing, and the status response says so.
Which languages are production-ready?: Japanese and English first, then Vietnamese, Indonesian, Thai, Filipino, Nepali and Chinese. About 85 languages are accepted in beta. GET /v1/languages returns the live table.
What happens to my media?: Input files are deleted as soon as the job finishes. Outputs are kept for 7 days behind short-lived signed URLs. Audio is processed by Google's paid Gemini API, which does not use it for training. Details in the Privacy Policy.
Can I bring my own timed text?: Yes. POST /v1/captions/format wraps your segments with the same Japanese and English rules and returns SRT, VTT and ASS synchronously, with no transcription and no Minutes.

How the APIs are offered

All APIs are listed on the RapidAPI marketplace. RapidAPI handles subscriptions, API keys, billing and invoices; YOSUGA K.K. operates the services themselves. Free tiers are available for evaluation. For volume pricing, enterprise terms or a data processing agreement, write to info@yosuga.io.

Captions that break the line where a human would.