Build with Traxlate.

Cancel a queued or in-progress job. Any credits that were reserved for the job are refunded immediately. Jobs that have already completed, failed, or been canceled return 409 Conflict.

Response

{
  "ok": true,
  "job_id": "01900000-0000-0000-0000-000000000001",
  "status": "canceled",
  "credits_refunded": 42
}

List the 50 most recent jobs for the authenticated key. Supports optional limit and offset query parameters.

{
  "jobs": [
    {
      "job_id": "018f…",
      "status": "done",
      "title": "Q1 Report EN→DE",
      "credits": 42,
      "created_at": "2026-05-12T09:00:00.000Z"
    }
  ],
  "total": 87
}

Returns the last 30 days of request and credit usage for the authenticated key. Days with no activity are included with zero values.

{
  "key_id": "key_abc123",
  "days": [
    { "day": "2026-05-12", "requests": 14, "credits_spent": 87 },
    { "day": "2026-05-11", "requests":  9, "credits_spent": 51 }
  ]
}

POST /api/v1/tts/synth

Render a single TTS clip. Returns the WAV bytes inline for text ≤ 1500 characters; longer text returns a jobId to poll.

Body

cURL — preset voice

curl https://traxlate.com/api/v1/tts/synth \
  -H "Authorization: Bearer tr_live_..." \
  -H "Content-Type: application/json" \
  -d '{"voiceId":"aria","targetLang":"en","text":"Hello, world."}' \
  --output hello.wav

cURL — cloned voice with an emotion preset

curl https://traxlate.com/api/v1/tts/synth \
  -H "Authorization: Bearer tr_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "voiceId":"aria",
    "cloneId":"7f5b...",
    "targetLang":"ko",
    "text":"こんにちは。",
    "emotionPreset":"warm"
  }' \
  --output ko_clone_warm.wav

POST /api/v1/tts/synth/stream

Same body as /synth, but the response is a chunked audio/wavstream whose RIFF size fields are 0xFFFFFFFF — players start playback at first-chunk arrival (~150–250 ms TTFT). Built for conversational agents.

POST /api/v1/tts/voices/clones

Upload a 3–20s reference audio clip. The engine validates, denoises, trims silence, detects gender. Returns a clone id usable on subsequent /synth calls. Storage is free; you only pay credits when you render with the clone. Per-user cap: 25 clones.

cURL

curl https://traxlate.com/api/v1/tts/voices/clones \
  -H "Authorization: Bearer tr_live_..." \
  -F "file=@reference.wav" \
  -F "name=My announcer" \
  -F "description=Warm neutral podcast host"

GET /api/v1/tts/voices

Returns the curated cast + the 31 supported languages. Public — no auth required.

Word-level editor

After a render, alignment + per-word DSP edits are available via /api/v1/tts/jobs/:id/words (alignment, lazy) and /api/v1/tts/jobs/:id/edit(apply pitch / speed / gain / pauses per word). Edits are free within the 12-hour retention window — refining your own render doesn't cost more credits.

POST /api/v1/subtitles/upload

Stream-upload the media file. Returns { video_id, duration_seconds, size_bytes }. The same video_id can be reused across dub + subtitle jobs — the engine caches the extracted audio per video id.

cURL

curl https://traxlate.com/api/v1/subtitles/upload \
  -H "Authorization: Bearer tr_live_..." \
  -H "x-filename: lecture.mp4" \
  -H "x-video-mime: video/mp4" \
  --data-binary @lecture.mp4

POST /api/v1/subtitles/create

Create a subtitle job. Omit targetLangs for caption-only (cheapest tier — source-language transcript only). Include one or more targets to also receive translated subtitle files.

Body

cURL — caption-only (cheapest)

curl https://traxlate.com/api/v1/subtitles/create \
  -H "Authorization: Bearer tr_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "videoId":"vid_abc...",
    "durationSeconds":900,
    "sourceLang":"auto"
  }'

cURL — caption + translate into FR & DE

curl https://traxlate.com/api/v1/subtitles/create \
  -H "Authorization: Bearer tr_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "videoId":"vid_abc...",
    "durationSeconds":900,
    "sourceLang":"en",
    "targetLangs":["fr","de"]
  }'

Returns { batchId, jobIds, mode, perChild, totalCredits }. All children share an engine job — polling any one child reflects the whole batch.

GET /api/v1/subtitles/:id/status

Poll job status. Subscribe to subtitle.done / subtitle.failedwebhooks instead if you don't want to poll. Subtitles are fast — a 30-minute video typically finishes in under a minute.

GET /api/v1/subtitles/:id/download/:fmt

Download a subtitle file. :fmt is one of srt, vtt, txt, bilingual_srt. Pass ?lang=xxto pick which language; omit for the job's default (source for caption-only, target for translate).

cURL

curl https://traxlate.com/api/v1/subtitles/abc.../download/srt?lang=fr \
  -H "Authorization: Bearer tr_live_..." \
  --output lecture_fr.srt

Free re-export from dubs

Every completed dub job exposes the same subtitle artifacts for free under /api/v1/dub/:id/srt, /vtt, /txt and /bilingual-srt. No extra credits — the dub pipeline already produced this data.