Inspiration
This project was inspired by the people that have Alzheimer's or Dementia and suffer constantly due to the consequences of it. Due to the memory loss effect associated with Alzheimer's, it may be difficult for one to remember the name of a certain person. To make their lives better, we decided to make this application to help those with dementia remember those names.
What it does
Our project is able to identify who is speaking by using Microsoft Azure's Speech Identification API that uses AI and an audio file. The file is recorded as monolithic .wav file and it has a sampling rate of 16k @ 16 bit PCM. The audio file is then compared to a dataset created with audio files of the Alzheimer's patient's relatives. This way, the input audio file is able to be compared to the dataset by Microsoft Azure's Speaker Identification services which decides to which person the input audio file most similarly sounds like.
How we built it
Our project was built with the use of an SDK for the Microsoft Speaker Recognition API. This SDK was written in Python and we heavily modified it to suit our needs for the project. We built it by using a python program to identify who is speaking and display the results after the program is run. The program would take in an audio sample as a base, which would then be read.
Challenges we ran into
We didn't have much prior experience with knowing how to use an SDK or an API with Python. This made it harder to get the ball rolling because it made the decision of finding the best speaker recognition API harder.
Accomplishments that we're proud of
We're proud of the fact that we were able to learn how to use an API and SDK just in time to finish our project. It let us know that most problems in computer science aren't as difficult or complicated as they seem.
What we learned
We learned how to work in python with Microsoft's API and how to utilize that to solve our problem. We were fairly new to working with python and was just introduced to the machine learning concept earlier this weekend.
What's next for Who is this?
We plan to further develop and refine the app and the program. Our program would have a harder time to discover a person provided there is background noise. We want to be able to void the background noise and make it easier for the person using the app to understand. Standalone objects are very useful in today's world, and we would like to make this project standalone (without the use of a computer), which would be more beneficial since it would be portable.
Log in or sign up for Devpost to join the conversation.