/transcribe endpoint

wss://api.medera.info/transcribe is the stateless dictation WebSocket for clinical editors. No session, no patient context, no persistence server-side — just real-time BH-tuned recognition with command-and-control. Built for the 5-minute window between encounters when the clinician is finishing a previous chart, signing refills, or paging a colleague.

Connection

wss://api.medera.info/transcribe?token=<jwt>

Configuration

{
  "type": "configure",
  "language": "en-US",
  "punctuation": "smart",
  "commands": ["new_paragraph", "next_section", "period", "comma", "delete_last"],
  "interim_results": true,
  "vocabulary": {
    "formulary": ["sertraline", "escitalopram", "lamotrigine"],
    "clinicians": ["Dr. Khalil"]
  }
}

Events

Event	Schema
`interim`	`{ text }`
`final`	`{ text }`
`command`	`{ command, args }`
`audio.event`	`{ kind }`
`error`	`{ code, message }`

Why “stateless”

No session is required. No transcript is persisted server-side. The endpoint is designed for clinical editor dictation where the editor owns the text. PHI never lands at rest from this endpoint.

Sample (JavaScript)

import { MederaClient } from '@medera/sdk';
const client = new MederaClient({ apiKey: process.env.MEDERA_API_KEY! });

const ws = await client.transcribe.connect({
  language: 'en-US',
  punctuation: 'smart',
  commands: ['new_paragraph', 'next_section', 'period'],
});

ws.on('interim', (e) => editor.previewText(e.text));
ws.on('final',   (e) => editor.commitText(e.text));
ws.on('command', (e) => editor.runCommand(e.command, e.args));

const mic = await navigator.mediaDevices.getUserMedia({ audio: true });
ws.streamAudio(mic);

What’s next

Commands

Build voice commands.

Smart punctuation

Punctuation modes.

/sessions — live capture

Stateful diarized session.

Microphones

Recommended hardware.

​Connection

​Configuration

​Events

​Why “stateless”

​Sample (JavaScript)

​What’s next