-
Notifications
You must be signed in to change notification settings - Fork 14
3. Registration
The registration process is straightforward.
- User enters their name and email
- User is prompted to say "Hey Nala" three times to store wakewords later.
- Various folders are made (wakewords, actions, queries, and baseline)
- Face is recognized and stored in ./baseline folder
- Databases are then made (registration.json, settings.json, and actions.json).
This is a future feature, but faces can be stored in the ./data/baseline folder for future use. This can be useful for authentication purposes later - for example, restricting Nala to registered users. It uses OpenCV and Haar cascades for face detection.
This would mean that your voice computer would have some sort of camera, as this is standard in most laptop computers.
Note that the new directories that are made include: wakewords, actions, queries, and baseline. These are all within the ./data/ folder.
Wakewords are where wakewords are stored if you want to train some of them later (e.g. ‘Hey Nala’). The actions folder is where Nala stores any data generated as a result of responding to user queries. The queries folder is where all the queries that a user makes are stored. Lastly, the baseline folder is where all the user data generated from the registration process is stored, along with any redundant backups of the database in case the master database fails.
Everything that Nala does is stored in freeform JSON databases. This makes it much easier to create complex databases and relationships. There are three main databases: registration.json, settings.json, and actions.json. These databases are elaborated upon in the DATABASES section of this wiki.