Skip to content

Quickstart

  1. Get a key

    Create an account at speechrouter.ai, verify your email (that unlocks the $2 welcome credit), and mint an API key in the console. Keys look like sk_sr_… — send them as a Bearer token.

  2. Transcribe a file

    Terminal window
    curl -s https://api.speechrouter.ai/v1/audio/transcriptions \
    -H "Authorization: Bearer $SPEECHROUTER_API_KEY" \
    -F model=deepgram/nova-3 \
    -F file=@meeting.wav
  3. Stream live audio

    import { SpeechRouter } from 'speechrouter'
    const sr = new SpeechRouter({ apiKey: process.env.SPEECHROUTER_API_KEY })
    const stream = sr.listen({
    model: 'soniox/stt-rt-v5',
    fallbacks: ['deepgram/nova-3'], // automatic mid-stream failover
    sampleRate: 16000,
    })
    stream.on('transcript', (t) => {
    if (t.is_final) console.log(t.text)
    })
    // feed 16-bit PCM chunks as they arrive from your mic / telephony leg
    stream.sendAudio(pcmChunk)
  4. Switch models

    Every model is a slug: provider/model. Change the slug, keep everything else:

    deepgram/nova-3 fastest general English
    soniox/stt-rt-v5 60+ languages, session-billed
    openai/gpt-4o-transcribe LLM-grade accuracy
    speechmatics/melia-1 56-language code-switching batch

    The full, live list with prices: GET /v1/models or the providers section.