moeru-ai/airi · warning

No audio input device selected

Error message

No audio input device selected

What it means

Hearing settings playground guard: setupAudioMonitoring() bails out with this warn (returning false) when selectedAudioInput is empty, before startStream() is attempted. The device list can be empty because enumerateDevices has not run yet, device labels/IDs are hidden until microphone permission is granted, the user chose nothing, or the machine has no input device.

Source

Thrown at packages/stage-pages/src/pages/settings/modules/hearing.vue:146

    const message = errorMessageFrom(cause) ?? t('settings.pages.modules.hearing.sections.section.playground.transcription-failed')
    finishError(metadata, message)
    error.value = message
  },
})

const isSpeechVolume = shallowRef(false) // Volume-based speaking detection
const isSpeech = computed(() => {
  if (useVADModel.value && loadedVAD.value) {
    return isSpeechVAD.value
  }

  return isSpeechVolume.value
})

async function setupAudioMonitoring() {
  try {
    if (!selectedAudioInput.value) {
      console.warn('No audio input device selected')
      return false
    }

    await stopAudioMonitoring()

    await startStream()
    if (!stream.value) {
      console.warn('No audio stream available')
      return false
    }

    if (supportsStreamInput.value) {
      // The Hearing pipeline owns speech segmentation and the provider session.
      // The page VAD below only drives the visualization for streaming providers.
      await transcribeForMediaStream(stream.value, {
        consumerId: hearingPlaygroundTranscriptionConsumerId,
        onSpeechEnd: finishStreaming,
        onTranscriptionUpdate: replaceStreamingText,

View on GitHub (pinned to 677329427f)

Solutions

1. Ensure askPermission() completed so navigator.mediaDevices.enumerateDevices() returns real device IDs
2. Pick a device in the Audio Input Device combobox, or auto-select the first audioinput once the list loads
3. Confirm the OS exposes a microphone to the browser (system privacy settings)
4. After unplugging a device, re-select a valid one - the empty selection will keep skipping setup

Example fix

// before
if (!selectedAudioInput.value) {
  console.warn('No audio input device selected')
  return false
}

// after - auto-pick the first available input
if (!selectedAudioInput.value) {
  const devices = await navigator.mediaDevices.enumerateDevices()
  const first = devices.find(d => d.kind === 'audioinput')
  if (!first)
    return false
  selectedAudioInput.value = first.deviceId
}

if (!selectedAudioInput.value) {
  const devices = await navigator.mediaDevices.enumerateDevices()
  const first = devices.find(d => d.kind === 'audioinput')
  if (!first)
    return false
  selectedAudioInput.value = first.deviceId
}
return await setupAudioMonitoring()

When it happens

Trigger: Opening the hearing settings module before the onMounted askPermission() resolves and devices are enumerated; permission never granted so enumerateDevices returns label-less entries; no microphone hardware; selectedAudioInput cleared after the chosen device disconnects.

Common situations: First visit racing the permission flow; revoked microphone permission; headless/VM environments without audio input; device unplugged between sessions leaving a stale empty selection.

Related errors

• Error stopping recognition instance:
• Invalid transcription model "${currentModel}" detected. Rese
• [Speech Pipeline] provider/voice/model changed mid-session,

AI-assisted analysis of moeru-ai/airi@677329427f (2026-08-18). Data as JSON: /api/errors/eed91d4d0c7410cb. Report an issue: GitHub.