Syntiant Corp. has introduced the TinyML Development Board, a single board computer aimed at both engineers and makers to design machine learning-powered smart products.
The introduction comes as single board computers are growing substantially in the market and have become a viable building block in all stages of new product development.
The board is powered by the Syntiant NDP101 neural decision processor, which the company launched last week, and can enable speech and sensor applications to run at under 140 to 100 microwatts, respectively. The NDP101 also has 20 times more throughput and 200 times more efficiency improvement compared to MCU-based boards.
The single board computer allows trained models to be downloaded via Edge Impulse through a micro-USB connection without any special hardware. It is also compatible with Arduino’s open-source platform.
Other features of the board include:
- Syntiant Core 1 neural network
- Up to 560 k parameters and 64 output classifications
- Arm Cortex-M0+ 32 bit low power ARM MCU
- 256 KB flash memory
- 32 KB host processor SRAM
- Five digital I/Os compatible with Arduino MKR series boards
- 2 MB on-board serial flash
- 48 MHz system clock
- Six-axis motion sensor
The TinyML development board allows users to sample raw data, build, download and deploy trained embedded machine learning modules to create applications that answer speech commands, wake word detection, acoustic event detection and other audio and sensor use cases.