Cartesia Ink vs OpenAI Speech to Text

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.

OpenAI Speech to Text

gpt-transcribe costs $0.0045 an audio minute, and the audio endpoints keep no abuse-monitoring logs or application state. Speaker labels, timestamps, subtitles and translation exist only on whisper-1 and gpt-4o-transcribe-diarize, which shut down on 26 February 2027 with no named replacement for those functions.

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.

OpenAI Speech to Text

1. Send gpt-transcribe to POST /v1/audio/transcriptions for recorded files. Use languages (a list), not language, and never send both.
2. Keep each upload at 25 MB or less. Split longer audio between sentences and pass the previous chunk's text in prompt.
3. For speaker labels send gpt-4o-transcribe-diarize with response_format=diarized_json and chunking_strategy=auto for audio over 30 seconds. Plan for its shutdown on 26 February 2027.
4. Word timestamps, srt, vtt and /v1/audio/translations need whisper-1, which cannot stream and shuts down on the same date.
5. On 429 or 503 wait at least Retry-After when present, then back off with jitter. Do not retry credit_balance_exhausted or spend-limit errors.