Transcription

Settings → Transcription sets how audio becomes text: in which language, and by which recogniser.

A conversation is transcribed when you ask for it in the Recordings window, or by itself if Process conversations automatically is on in Processing. A recogniser on your own machine costs nothing to run; one in the cloud charges by the minute of audio.

Language

Language is a two-letter language code as in ISO 639-1 (en, de, es, fr, sr…). Leave it empty and the recogniser decides — that is right unless your calls are in a language it keeps mishearing.

Recognisers

A recogniser is a speech-to-text service the phone sends audio to. Press Add to add one; the Test button of the form checks that the service really answers. Each is listed with its name and, under it, the model and the address of its service. In the picture there are four:

Name	Model	Address
X.ai	(empty: the service default)	https://api.x.ai/v1
ElevenLabs	scribe_v2	https://api.elevenlabs.io/v1
Deepgram	nova-2	https://api.deepgram.com/v1
OpenAI	gpt-4o-transcribe	https://api.openai.com/v1/

The one marked default at the right of its row (X.ai in the picture) is the one used when you do not choose another. You can keep several. The drop-down above a transcript in the Recordings window lists the transcripts made by each recogniser.

The model may be left empty. The service then uses its own default.

Which model to choose

The table lists the speech-to-text models of the four services in the picture. The models in bold are the ones set up in the picture. For the X.ai recogniser the model is empty, so the service default, grok-voice-transcribe-2.0, is the one used.

Service and address	Model	What it is for
OpenAI https://api.openai.com/v1	gpt-transcribe	The model OpenAI recommends for recorded speech in its original language.
	gpt-4o-transcribe	General-purpose transcription.
	gpt-4o-mini-transcribe	A lighter, cheaper variant of the above.
	gpt-4o-transcribe-diarize	Labels who speaks when. Use it only if you need that.
	whisper-1	The older Whisper model, kept for special uses such as word timestamps and subtitles.
ElevenLabs https://api.elevenlabs.io/v1	scribe_v2	General-purpose transcription in 90+ languages, with speaker separation.
	scribe_v2_medical	The same, tuned for clinical audio.
	scribe_v1	The first generation; deprecated, use scribe_v2.
Deepgram https://api.deepgram.com/v1	nova-3	Deepgram's best general-purpose model, for meetings, noisy and multilingual audio.
	nova-2	The previous generation; keep it for languages nova-3 does not support yet.
	enhanced	An older tier with lower error rates than base.
	base	The oldest tier, for large volumes.
	whisper	Whisper, run by Deepgram.
X.ai https://api.x.ai/v1	grok-voice-transcribe-2.0	The default; 25 languages.
	grok-voice-transcribe-1.0	Deprecated: the service sends it to 2.0.

Things worth knowing before you choose:

• File size. OpenAI takes files up to 25 MB; X.ai up to 500 MB. A long conversation can be larger than a cloud service accepts.
• Price. Cloud services charge by the minute of audio, and the rates differ by model and change; read them on the service's own page before you switch.
• Languages. Every service has its own list; check yours, and set the Language code if the recogniser guesses wrongly.
• Real-time models such as scribe_v2_realtime or Deepgram's flux are made for live streams and are not in the table: the phone transcribes finished recordings.

The list of a service's models changes often. If a model you want is missing here, the service's own documentation has the current list — OpenAI, ElevenLabs, Deepgram, X.ai — the Model is the name exactly as the service gives it.

Your own models

A recogniser does not have to be a cloud service. The phone can use any model that is served through the OpenAI-compatible API — the POST /v1/audio/transcriptions interface — whether it runs locally on your computer or on a server of your own. The audio never leaves your premises, nothing is charged by the minute, and there is no limit on volume.

To add one, press Add and give:

• the address of the server, up to and including /v1, for example http://localhost:8000/v1 for the computer itself or http://asr.local:8080/v1 for a server in your network;
• the model name exactly as the server lists it, for example openai/whisper-large-v3-turbo.

What can be used

The usual choice is Whisper, OpenAI's open speech recognition model. It is free to use, understands about a hundred languages, and comes in several sizes: a small model runs on an ordinary computer, the large ones are noticeably more accurate and are best given a graphics card.

Model	Notes
whisper-large-v3	The most accurate Whisper. For a server with a GPU.
openai/whisper-large-v3-turbo	A faster version of large-v3 with a small loss of accuracy.
Systran/faster-whisper-large-v3	large-v3 converted for the faster-whisper engine; quicker, and lighter on memory.
medium, small, base	Smaller Whisper models, for a computer without a graphics card.

Whisper is the model that these servers are built around. Some of them can also serve other speech recognition models, such as NVIDIA Parakeet.

Servers that offer the OpenAI-compatible API

The model has to be run by a server that offers the OpenAI-compatible /v1/audio/transcriptions endpoint. These do:

Server	What it is
vLLM	A high-performance model server. Serves Whisper at http://localhost:8000/v1 once started.
Speaches	A server for speech models, "Ollama for speech", built on faster-whisper. Loads a model when it is first asked for.
whisper.cpp	Runs Whisper efficiently on a CPU, including Apple silicon. Its whisper-server is started with --inference-path /v1/audio/transcriptions.
LocalAI	A drop-in OpenAI replacement that runs models locally.

Any other server that offers the same endpoint works in the same way. If a server needs a key, enter it as for a cloud service.

Before you rely on a server, make a test recording and look at the transcript in the Recordings window: a conversation in a language the model knows poorly shows it at once.