[Interactive Devices] Be Nice – Final Project Documentation

Be Nice – A Speech Recognition Device
by Putri Dina Andyana

Project Overview
As millennials get influenced by media, using insensitive words in their everyday conversations has become something they do unconsciously. The words definitely affect our vocabulary and more importantly, offend some people who regard them as an insult. There are definitely other phrases that we can use.

Be Nice is a speech recognition device which translates insensitive words into alternative phrases which are less offensive to the other party. With this project, the end goal is to evoke a sense of self-awareness to whatever that is coming out of our lips.

How It Works?
Audio will be captured and amplified using the electret microphone which sends them over to an Arduino Due. It digitalises via Analog-to-Digital Converter (ADC). A software called ‘BitVoicer Server’ processes this audio and recognise speech to predefined commands.

Once it identifies a command, it sends back to Arduino and queue it into a BVSSpeaker library to reproduce the audio via Digital-to-Analog Converter (DAC). A Mono Audio Amp amplifies the audio through and 8 Ohm speaker. A row of LED lights will indicate how insensitive that word is (yellow being the least and red the most).

Project Process Timeline
[101017] – Proposal & First Prototype
[181017] – Initial Wiring

[201017] – Input Speech & Output LED
[231017] – Speaker Test & Output Synthesizer
[061117] – Soldering & Final Paper Prototype
[141117] – Laser Cutting & Assembling
[161117] – Finalizing
[171117] – Final Project Documentation

Difficulties Encountered
< Technical >
Speech recognition. That pretty much explains most of my problems. Although the technology is not there yet, I still decided to take on the challenge of using speech recognition in this project.

Although the Arduino code worked perfectly fine, the main problem came from BitVoicer Server where at times it was not able to recognise my voice (or any other voice actually).

The location of the user also played a part as bigger and wider room with a lot of noise would decrease the sensitivity of the microphone. Therefore, I had to test the device in my room most of the time and ‘calibrate’ the Minimum Audio Level and Confidence Level whenever I moved the device another location (such as the IM Lab).

< Packaging >
I had not much problem when it came to the physical container of the electronics except for my big mistake of spray painting the acrylic first before glueing them together. The acrylic glue was too strong that it also blended my paint away. Hence, visible stains could be seen.

Further Development
When the technology for speech recognition advances further, it can eliminate the issues of identifying the exact words spoken by the user. The device can be more compact if an alternative circuit board or programming method is used.

Currently, the device has to be plugged in to my computer which contains the BitVoicer Server. There is an option of using WiFi to allow the software to communicate with the Arduino. However, it is too complicated for me to implement it for this project. If that is possible, it can be a portable device, only powered by a battery pack.

Published by

Dina

Believes in creating works that someone can not only see or touch but be part of, to be within them.

Leave a Reply