Beta — public API live

Transcripts and AI syntheses for any source.

Send a YouTube URL, a podcast feed or an audio file. Get back a structured transcript with timecodes and an AI-generated synthesis you can embed in your product. Pay only for what you ingest.

Create your first token Read the quick start

Any source

YouTube, podcast RSS, audio URL, web articles. Same endpoint.

REST + JSON

No SDK required. curl, Python, Node, anything that speaks HTTP.

Async pipeline

Submit, poll for status. We do the heavy LLM work in the background.

Per-token tracking

Multiple tokens per user. Rotate, revoke, attribute cost — your call.

Quick start

Three steps to get a synthesis: create a token, submit a source, poll until it's ready.

1. Create a token

Go to /settings/tokens, click Créer, copy the secret. Tokens start with synz_live_.

2. Submit a source

curl -X POST https://synthiz.com/v1/sources \
  -H "Authorization: Bearer synz_live_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
    "source_type": "youtube"
  }'

# → 202 Accepted
# {
#   "source_id": "abc-123",
#   "memoire_id": "...",
#   "workflow_id": "...",
#   "status": "accepted"
# }

3. Poll for status

curl https://synthiz.com/v1/sources/abc-123 \
  -H "Authorization: Bearer synz_live_xxx"

# → 200 OK
# {
#   "source_id": "abc-123",
#   "status": "synthesized",        // or "pending", "transcribed"
#   "has_transcript": true,
#   "has_synthesis": true,
#   ...
# }

Poll every 5–10 seconds. Most sources are ready in 30 s – 5 min depending on length.

4. Fetch the transcript

curl https://synthiz.com/v1/sources/abc-123/transcript \
  -H "Authorization: Bearer synz_live_xxx"

# → 200 OK
# {
#   "source_id": "abc-123",
#   "text": "first segment\nsecond segment...",
#   "segments": [
#     { "position": 0, "text": "...", "start_seconds": 0, "end_seconds": 4.2 },
#     ...
#   ]
# }

Endpoints

Method	Path	What it does
POST	/v1/sources	Submit a URL for ingestion. Returns a source_id.
GET	/v1/sources/{id}	Status + metadata. Poll this until status=synthesized.
GET	/v1/sources/{id}/transcript	Full transcript with timecodes.
GET	/v1/sources/{id}/synthesis	AI synthesis (title, summary, concepts). Coming soon.
GET	/v1/me/usage	Your consumption breakdown (cookie-auth, used by /settings).

Supported source types

youtube — any youtube.com / youtu.be URL with captions or audio
podcast — RSS feed or episode URL (Apple Podcasts, Spotify…)
audio_url — direct link to an MP3/M4A/WAV/OGG file
webpage — any HTML article URL

Multipart upload of local files is on the roadmap. For now, host your file somewhere reachable (S3, CDN…) and pass its URL.

Limits

Per-token rate limit: 60 requests / minute
Global cap: 1 000 requests / minute across all tokens
HTTP 429 on overage with a Retry-After header
Ingestion time: typically 30 s – 5 min depending on source length

Pricing

Pricing is being finalised. Beta accounts can ingest freely up to a generous monthly cap — we'll notify you well before any limit kicks in.

Cost is tracked per-source and per-token under /settings/usage. You always see in real time what an ingestion costs Synthiz, so we can price transparently.

Need help?

The API is in beta — feedback and bug reports go a long way. Drop us a line through the in-app feedback button or send a mail to hello@synthiz.com.