Cartesia Ink vs ElevenLabs Scribe Speech to Text API

Cartesia Ink

Ink 2 streams transcripts with turn detection built into the model, documented by an AsyncAPI file with typed ranges and structured errors. The terms, last revised 14 June 2024, let Cartesia train on inputs and outputs unless the customer opts out, and zero data retention is an Enterprise setting. Ink 2 has no batch endpoint and no diarisation.

ElevenLabs Scribe Speech to Text API

$0.22 an hour for batch with 90+ languages and diarisation to 32 speakers. Audio may be used for training unless the account opts out, and the opt-out isn't retroactive.

Cartesia Ink

1. Use wss://api.cartesia.ai/stt/turns/websocket with model=ink-2, encoding, sample_rate and cartesia_version=2026-08-14. All four are required.
2. Send raw mono audio in chunks of about 100 ms at the speed it was spoken. Pushing a whole file into the socket can return an internal server error.
3. Read the final text from turn.end only. transcript is cumulative within a turn, so joining turn.update events duplicates text.
4. Send {"type": "close"} after the last audio and keep reading until the server closes the socket, or the buffered tail is lost.
5. Check encoding and sample_rate against the source before sending. The docs say the server might not return an error when they are wrong.

ElevenLabs Scribe Speech to Text API

1. Create a key restricted to speech-to-text with a credit quota and an expiry
2. Use source_url for hosted files instead of downloading and re-uploading
3. Set webhook=true for long files so the call doesn't block
4. Back off exponentially on any of the three 429 codes
5. Opt out under Data use before sending customer audio. It only covers later uploads