Voice Lexicon API

API which generates short audio messages in two languages in order to speed up your learning of a new language

Comment

Inspiration

Learning new vocabulary is a demanding process and it is even more daunting for those who suffer from learning disability, according to Dyslexia Center Utah 20% of the elementary school population is struggling with reading. One of the way to increase the efficiency of learning is the multi-sensory approach – to use more senses like audio. We know on our own as one of the members is learning the third language with diagnosed dyslexia. That is why we want to improve the accessibility of vocabulary in audio format and make it available on Alexa.

What it does

Voice Lexicon creates an audio file based on providing vocabulary on the website and enables a user to play this file on any Amazon Alexa with our skill and special two-word phrase. This way, learners can seamlessly share their vocabulary with anyone around the world.

How we built it

We used Amazon Polly text-to-speech technology to generate a multi-language audio file from your vocabulary and make it available through Voice Lexicon skill and provided phrase on any of the 50 million available Alexa devices. Backend runs on AWS Lambda to keep costs down.

Challenges we ran into

One of the challenging things was to design the way to generate phrases that will be used for playing vocabulary. What helped us is bitcoin Wordlists of English phrases that are used as a backup of private keys. Those words are pre-sorted for such use case.

Accomplishments that we're proud of

Being able to combine web with Alexa in a meaningful way.

What we learned

AWS Lambda can be useful tool to keep cost of project low

What's next for Voice Lexicon API

We can significantly improve remembrance of the phrases by structuring them with and avoiding abstract words that are harder to remember - but or know we do not know if this is a problem. Also, we would like to create our vocabulary for people who start with learning and do not have their word list yet. The biggest improvement would be creating a testing function, but there are several challenges like changing the language of input or getting the raw audio file.

Built With

Share this project:

Updates