Earlier this year, we wrote about Adafruit porting TensorFlow Lite to the Arduino IDE. At the time, they demonstrated voice recognition using their PyGamer board and an add-on microphone. Adafruit’s newest board is the EdgeBadge, which is about the size of a credit card (~86 mm x ~54 mm). It is a follow-on to the PyGamer, which means it comes with an Arm Cortex-M4F running at 120 MHz with 192 KB of RAM and 512 KB of flash. The goal is to run TensorFlow Lite on this small development platform. For that reason, the EdgeBadge also includes a 2 MB QPSI flash. You can use the extra flash for stuff like sounds, images, or TensorFlow Lite files.
EdgeBadge running TensorFlow Lite
Other than the sleek looking dark design, the primary difference from the PyBadge is the on-board PDM microphone. This upgrade is not a surprise since Adafruit’s first TensorFlow Lite demonstration was voice recognition. By the way, PDM means pulse density modulation. The microphones in most mobile devices use the same method to deliver data back to the signal processor. (It is similar to PCM.)
If you are familiar with the PyBadge, then you already know the specs for the EdgeBadge. If not, in addition to the ATSAMD51, the board contains a 1.8″ 160×128 Color TFT, 8 buttons in a game controller layout, a LiPo battery charging circuit, and two female header strips compatible with FeatherWings.
Adafruit EdgeBadge supports FeatherWings
Whether you want to join #badgelife at your next tech conference or you want an inexpensive platform to get started with machine learning, the EdgeBadge has you covered. For more information, check out the EdgeBadge product page on Adafruit.