Linden 1
Linden 1 was built for voice agents. It first became available through LiveKit Inference on 27 August 2026, before Speechmatics launched its direct Agent STT interface.
Short replies and spoken identifiers
The development account highlights short confirmations, accented telephone speech and alphanumeric strings. Those inputs can break an agent even when an ordinary transcription benchmark looks good. Linden’s work on account numbers and addresses targets that problem.Speechmatics: Agent STT powered by Linden 1
Agent STT returns speaker-labelled segments and turn events through a WebSocket connection. It can route to a nearby region or use a specified region. The output is recognition data for an application; Linden does not generate the agent’s reply.Speechmatics documentation: Agent STT
Practical guidance. A useful evaluation includes one-word replies and complete spoken identifiers. A single incorrect character can stop an account lookup even when the rest of the sentence is accurate.Speechmatics: Agent STT powered by Linden 1
One model behind several integrations
LiveKit Inference handles its own routing, billing and turn detection around Linden. LiveKit access and the direct Speechmatics API are distribution routes for the same named model.Speechmatics: Linden arrives in LiveKit Inference
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | API |
| Model identifiers | linden-1Linden |
| Direct API launch | 17 September 2026Speechmatics: Agent STT powered by Linden 1 |
| Service route | Agent STT; SaaS onlySpeechmatics documentation: Agent STT |
| Custom vocabulary | Up to 1,000 added termsSpeechmatics: Agent STT powered by Linden 1 |
Related models
- Melia 1
Speechmatics · Melia
Multilingual speech recognition model that follows language changes within a recording.
Launch date API · Provider product - Enhanced
Speechmatics · Enhanced
General speech recognition model that prioritizes accuracy across supported language packs.
Launch date not confirmedAPI · Provider product - Text-to-Speech Preview
Speechmatics · Speechmatics TTS
English speech synthesis models with streamed audio and four selectable voices.
Launch date not confirmedPreviewAPI · Provider product