The Home Assistant Voice Preview Edition is a voice assistant hardware which seamlessly integrates with Home Assistant. Its audio processor and dual microphones allow it to hear your commands and control devices in Home Assistant similar to Amazon Echo.
The Getting Started Guide is here https://support.nabucasa.com/hc/en-us/ar...ew-Edition
one of the many Youtube videos on the setup and usage is https://www.youtube.com/watch?v=vceoL59pmkk
Setting up is easy. However note that if you use Duckdns with Lets Encrypt so that access locally and remotely by by https, Voice PE will not work as it requires access by http for setup and operation.
You can select a preferred pipeline for processing the voice. This is done at setup but you can change it in Settings > Voice Assistants

A good description of these options are in the video https://www.youtube.com/watch?v=zhlIaBG3Ldo at around 2 minutes.
My experience is the performance of the speech recognition is far better with Nabu Casa Cloud (which requires a subscription) compared to the local processing options, though this may be due to my accent - It may work better for european and american voices. There may be ways to tune the speech recognition engines but I have not explored them. The Cloud processing option is also significantly faster than the local processing.

The Getting Started Guide is here https://support.nabucasa.com/hc/en-us/ar...ew-Edition
one of the many Youtube videos on the setup and usage is https://www.youtube.com/watch?v=vceoL59pmkk
Setting up is easy. However note that if you use Duckdns with Lets Encrypt so that access locally and remotely by by https, Voice PE will not work as it requires access by http for setup and operation.
You can select a preferred pipeline for processing the voice. This is done at setup but you can change it in Settings > Voice Assistants

A good description of these options are in the video https://www.youtube.com/watch?v=zhlIaBG3Ldo at around 2 minutes.
My experience is the performance of the speech recognition is far better with Nabu Casa Cloud (which requires a subscription) compared to the local processing options, though this may be due to my accent - It may work better for european and american voices. There may be ways to tune the speech recognition engines but I have not explored them. The Cloud processing option is also significantly faster than the local processing.





