let's talk about the cabin of a vehicle, ... Mercedes... recognize the driver, recognize voice, recognize gestures, ...
We had confirmation of in-cabin key word spotting with EQXX with a suggestion that there may be more ...
- now Rob is talking about driver recognition, gesture recognition, and voice recognition (is that individual voices as opposed to general word spotting?) in the context of in-cabin monitoring and, by-the-way, this could be in, for example, Mercedes ... and still being developed, just as MB said the other day in the context of announcing a development partner (they have already announced Nvidia and Qualcomm) in the not-too-distant future.
Well the Akida silicon has been developed, but what else needs to be "developed" is the data models for speech, gestures, etc, and while there are some general purpose models available for CNN software, the specific in-cabin models are a horse of a different colour (remind me to tell you about Green Horse). And, if Akida is to be used, the models should be in SNN format. I don't now whether they can be developed in SNN directly (it is possible in theory), but it may be faster to develop CNN models and convert to SNN.