These are the settings that decide how the device listens and speaks: the
wake word, how it tells speech from silence, the plugins that turn speech into text and text into speech, the
order it tries to understand a command, and the language. Every change is written into your own layer of
mycroft.conf, the same file the Settings page's Advanced tab edits — this page just gives each
setting a plain picker.
The wake word, the speech-to-text engine and the silence detector apply as soon as you save. The order the device tries to understand a command needs a restart, and so does changing the voice. If something does not change, restart the OVOS services.
Your changes are stored in your own layer of this file:
Wake word
The name the device listens for before it pays attention to you. Only wake
words already set up in the hotwords section can be chosen here.
No wake words are set up on this device yet. Add one in the Advanced tab of Settings, or install a wake-word plugin on the Plugins page.
Listening (voice activity)
The plugin that decides where your speech starts and stops, once the wake word has fired.
No voice-activity plugins are installed. Install one on the Plugins page; the current value is kept until you change it.
Speech to text
The plugin that turns what you say into text. A fallback is tried only when the primary one fails.
No speech-to-text plugins are installed. Install one on the Plugins page.
Text to speech
The plugin that speaks the device's replies. A fallback is tried only when the primary one fails.
No text-to-speech plugins are installed. Install one on the Plugins page.
This plays a sample on the device using the voice that is in use now. To hear a new choice, save it first, then press it again.
Advanced — pipeline order and transformers
Most people never need to change these. They decide how the device works out what you mean.
Pipeline order
When you speak, the device tries each of these in order and stops at the first one that understands you. Move an entry up to have it tried sooner.
Transformers
Utterance transformers clean up or change what you said before the device tries to understand it. Metadata transformers add extra information to a request before it is handled. Turn one on or off below.
For all six transformer chains, their order, and each plugin's own settings, use the Transformers page.
Utterance transformers
No utterance-transformer plugins are installed.
Metadata transformers
No metadata-transformer plugins are installed.
Language
The language the device speaks and listens in, for example
en-us or pt-pt. Extra languages can be added below; the device also tries to understand
those, alongside the main one.
To have the device answer back in an extra language, not just understand it, turn on bidirectional translation on the Transformers page.