David Pine - Magic mirror on the wall, who is the fairest one of all?

Play David Pine - Magic mirror on the wall, who is the fairest one of all?
Sign in to queue


David Pine shows us his magic mirror, a screen in a mirror, that can display useful information such as his schedule for the day, the weather forecast, and much more. The mirror uses a Raspberry Pi 3 running Windows 10 IoT Core, and runs a custom, open source UWP application. It also has a camera, microphone, and sound bar, enabling voice-based interactions.

David's blog post: https://ievangelist.github.io/blog/building-a-magic-mirror/

The magic mirror repo: https://github.com/IEvangelist/Mirror



The Discussion

  • User profile image
    Larry Tindell

    Pls Thank Kendra for the "Heads Up" on this "Magic Mirror" "Windows 10 IoT Core" Prezy and
    Discussion. I had started work on a smaller mirror using an old -- working but 'OS-Defunct' --
    MS Win laptop. I particularly like your inclusion of voice recognition ( and response ). Thx !
    == YES -- "We are not just developing software, We are developing ( a) CULTURE !" ========

  • User profile image
    Larry Tindell

    IMHO -- A "Magic Mirror" Does NOT REQUIRE Full Voice "NLP" (nor voice response ) -- the primary
    objective is simply to be "hands-free", e.g. when shaving or applying make-up [ I do the former but
    could really use help with the latter :) ]. I programmed a voice response unit for a govy agency that
    uses an (unfortunately, proprietary) voice (limited) speech recog / command recognition lib ('C') to
    both setup a user's commands and to then -- with some 'training' -- to respond to these patterns.
    Have U checked "System.Speech" in dotNet ? SAPI is NOT under 'U' dotNetCore. Try a Kinect I/F ? ? ?

  • User profile image
    Larry Tindell

    Since I have not seen a comment or response from anyone (timezones?), I will "finally" mention that
    "CMU Sphinx" has already been tested on Raspberry Pi and "pocketsphinx" made to work with C# ...

  • User profile image

    @Larry Tindell: thank you for all your comments. I apologize for the delay in my response.
    The voice recognition is using the UWP speech libraries. They are fairly easy to consume and integrate with the mirror, UWP Speech Interactions. I did however struggle justifying the amount of processing power and network I/O that was required to do some of the more intensive work, i.e.; using Cognitive Services. There was mention during the interview of the potential to include Kinect, which I really like. I realized that I didn't get to actually demo any of the code or how to debug the app. One of the advantages is the ability to use Visual Studio and the Windows Device Portal.

Add Your 2 Cents