Streaming STT

import { STT } from '@runanywhere/web'

// Create a streaming session
const session = STT.createStreamingSession()

// Feed audio chunks as they arrive from the microphone
function onAudioChunk(samples: Float32Array) {
  session.acceptWaveform(samples)

  // Check for partial results
  const result = session.getResult()
  if (result.text) {
    console.log('Partial:', result.text)
  }
}

// When done speaking
session.inputFinished()
const finalResult = session.getResult()
console.log('Final:', finalResult.text)

// Clean up
session.destroy()

STT.createStreamingSession(options?: STTTranscribeOptions): STTStreamingSession

interface STTStreamingSession {
  /** Feed audio samples into the session */
  acceptWaveform(samples: Float32Array, sampleRate?: number): void

  /** Signal that no more audio will be provided */
  inputFinished(): void

  /** Get the current transcription result */
  getResult(): { text: string; isEndpoint: boolean }

  /** Reset the session for a new utterance */
  reset(): void

  /** Release all resources */
  destroy(): void
}

import { STT, AudioCapture } from '@runanywhere/web'

const capture = new AudioCapture()
const session = STT.createStreamingSession()

// Feed microphone audio into the streaming session
capture.onAudioChunk((samples) => {
  session.acceptWaveform(samples, 16000)

  const result = session.getResult()
  if (result.text) {
    document.getElementById('transcript').textContent = result.text
  }

  if (result.isEndpoint) {
    console.log('Endpoint detected:', result.text)
    session.reset() // Ready for next utterance
  }
})

// Start capturing
await capture.start({ sampleRate: 16000 })

// Stop when done
// capture.stop()
// session.destroy()