Speech recognition with an Arduino Nano

By on August 5, 2021
Pin It

Peter Balch, reading an IEEE report on speech recognition from the late 1970s, wondered if, with a normal Arduino Nano, it was possible to obtain the results of a computer of the time.

Comparing the two tools a Nano has 2 KB of RAM, 32 KB of program ROM and runs at about 10 MIPS (depending on the instruction mix); the type of computer people used at the time ran from 0.5 to 8 MIPS and had 2K to 32K of memory split between program and data. Is it then possible to get any results?

For this project, you will need an Arduino Nano (or Uno or Mini or similar as long as you use a 16 MHz ATmega328), a microphone and an amplifier for the microphone. In this case we use the MAX9814 microphone amplifier because it has an ACG.

The final result is, under ideal conditions, 90% to 95% correct recognition, which is roughly what people were getting in the 1970s. As for the overall software system, the training is performed on a PC but the trained system is performed entirely on the Arduino. Below is a summary image.

More info

About Bianchi Gabriele

Leave a Reply

Your email address will not be published.