# Speech recognition

Convert audio to text via BitrixGPT 5.6 Transcribe on Bitrix24 infrastructure — no BYOK required. Neither model is offered on the international platform yet: transcription is not available there, and every request is rejected before recognition — see the Models table on the transcription page. Transcription counts against your Bitrix24 account's AI quota by audio duration: within your plan's quota, nothing is charged to the balance; beyond the quota, usage is charged to your account's money balance at the model's base price.

Whisper Large v3 Turbo is being retired. From 16 November 2026, requests with `bitrix/deepdml/faster-whisper-large-v3-turbo-ct2` and requests without `model` are served by BitrixGPT 5.6 Transcribe. The old ID remains accepted and remains the default when `model` is omitted. For new integrations, explicitly set `model=bitrix/bitrixgpt-5.6-transcribe`.

Scope: `vibe:ai`

## Operations

- [Transcribe audio](./audio/transcriptions.md) — `POST /v1/audio/transcriptions`

## Typical scenario

1. Obtain an audio recording (a call, a voice message, an audio attachment).
2. Pass the file to [`POST /v1/audio/transcriptions`](./audio/transcriptions.md) — get the transcribed text.
3. Pass the text to a [chat completion](/docs/ai/chat/completions) for classification, summarization, or entity extraction.

## See also

- [AI Router](/docs/ai)
- [Chat completions](/docs/ai/chat)