Troubleshooting

July 31, 2026 · View on GitHub

Wakeword triggers but nothing happens (no transcription, no spoken response)

Most likely cause: the hivemind-core server is missing the hivemind-audio-binary-protocol plugin.

hivemind-core does not handle the recognizer_loop:b64_transcribe or speak:b64_audio messages. The relay sends audio to the server and waits up to 20 seconds for a transcription response; when none arrives it logs:

Timeout waiting for STT transcriptions

and returns an empty string. No intent fires and no TTS audio is sent back.

Fix: ensure hivemind-core has the hivemind-audio-binary-protocol plugin installed. Alternatively, run hivemind-core together with ovos-audio and ovos-dinkum-listener on the same machine to provide the same capabilities.

No audio captured / microphone not found

Error opening ALSA device

or the relay starts but never triggers on speech.

Check that the microphone is visible to ALSA:

arecord -l

If no devices are listed, the microphone is not recognised by the OS. Check USB/driver support.

If the wrong device is selected, specify it in ~/.config/mycroft/mycroft.conf:

{
  "microphone": {
    "module": "ovos-microphone-plugin-alsa",
    "device_name": "plughw:1,0"
  }
}

Try recording a short clip to confirm the device works:

arecord -D plughw:1,0 -d 3 test.wav && aplay test.wav

TLS / self-signed certificate error

ssl.SSLCertVerificationError: [SSL: CERTIFICATE_VERIFY_FAILED]

The server is using a self-signed certificate. Pass --selfsigned:

hivemind-voice-relay --selfsigned

Wakeword never triggers

Confirm you are using the correct wake word. The default is "hey mycroft". Check listener.wake_word in ~/.config/mycroft/mycroft.conf.
Confirm the wake word plugin is installed:
```
pip show ovos-ww-plugin-precise-lite
```
Test VAD separately — VAD silence incorrectly rejecting speech will prevent frames from reaching the wakeword detector. Try switching VAD to ovos-vad-plugin-webrtcvad to rule out a Silero model issue:
```
pip install ovos-vad-plugin-webrtcvad
```
```
{ "listener": { "VAD": { "module": "ovos-vad-plugin-webrtcvad" } } }
```
Increase microphone gain at the OS level:
```
alsamixer
```

NodeIdentity not set error

RuntimeError: NodeIdentity not set, please pass key/password/host or call 'hivemind-client set-identity'

No credentials are stored and none were passed on the command line. Run:

hivemind-client set-identity --key YOUR_KEY --password YOUR_PASSWORD --host wss://your-host

Or pass all three flags directly:

hivemind-voice-relay --key YOUR_KEY --password YOUR_PASSWORD --host wss://your-host

TTS audio received but no sound plays

Check that the default ALSA output device is correct:

aplay -l
aplay /usr/share/sounds/alsa/Front_Center.wav

If the wrong card is selected, set defaults in ~/.asoundrc:
```
defaults.pcm.card 1
defaults.ctl.card 1
```
Check volume is not muted:
```
alsamixer
```

Connection drops and does not reconnect

The hivemind-bus-client library handles reconnection internally. If the service exits on disconnect rather than reconnecting, the systemd Restart=on-failure directive will relaunch it. Check the service unit includes:

Restart=on-failure
RestartSec=5

PHAL not available

PHAL is not available

This is an informational message, not an error. PHAL (platform hardware abstraction layer) is optional. If you need platform-specific hardware support (e.g. LEDs on a Mark 1), install:

pip install ovos-PHAL

Otherwise the message can be ignored.

← Development · Home