Closed stintel closed 9 months ago
This idea seem quite reasonable, we will add one.
The friendly name was defined in _MODELINFO file.
We add esp_srmodel_get_wake_words function to parse friendly name from model_info
. If there are multiple wake words in one model, the wake words will be joined by ";".
You can use it like this:
srmodel_list_t *models = esp_srmodel_init("model");
char *model_name = esp_srmodel_filter(models, ESP_WN_PREFIX, NULL);
char *wake_words = wake_words = esp_srmodel_get_wake_words(models, model_name );
In Willow, we set a wake help text on the display to inform the user which wake word to use, e.g.
Say 'Alexa' to start
. We currently hardcode the friendly name in the C code. This works fine for the 3 wake words we currently support. In the future, we would like to offer our users a way to flash new models as they become available, without having to update the Willow C code, so this would no longer work.Would you consider adding a field to the WakeNet models for such a friendly name, so we can remove the hardcoded friendly names from our C code? Thanks!