ESP32 Forum

The current firmware v0.2.1 for the ESP-BOX uses wakenet_model hiesp8 which does not work. Replacing it with wakenet_model alexa8 does work.

could you elaborate more?
btw, it's "hi E-S-P"... for reference: https://youtu.be/n0jH1Brmee8

See this issue https://github.com/espressif/esp-box/issues/9 for more info.

I even tried playing the video you reference to the ESP-BOX with no result.

Updating ESP-BOX using https://github.com/espressif/esp-box/tr ... ctory_demo built with the alexa wake word model and playing it this video https://www.youtube.com/watch?v=jf6G5KKGvSo works fine even when noisy.
The multinet_model also works fine.

Hi Pibbotley
I have checked the firmware v0.2.1 for the ESP-BOX. We can wake up by "Hi,ESP"
When I play the video https://youtu.be/n0jH1Brmee8, the ESP-BOX can not be woken up.
Can you try the attached sample to see if the sample can work.

The ESP-BOX does respond to the sample you provided, and I can get the box to respond if I use my voice with a similar accent, but it is not a natural english speaking intonation.

The english hi esp firmware is very bad but someone has kindly used the alexa model and it works fine.

https://github.com/s60sc/ESP-BOX_firmware_Alexa

We have updated a new "Hi,ESP" model by adding more training samples.
You can pull the latest esp-sr (release/v1.0) to try it. https://github.com/espressif/esp-sr/tree/release/v1.0

New hiesp8 model is a significant improvement

I have this same problem with not responding to hi ESP wake word, as it works in youtube video's its obvious the wake word model is bad or the engine code is wrong.

basically the v021 en demo is broken, a broken demo is a little embarrassing especially when shipped with it.

Kindly remind: we have updated the new application firmware, which updates the speech recognition model, add features of continuous identification, the GUI interface is compatible with current two box series development boards. You can download and experience it. https://github.com/espressif/esp-box