We thought about the problems faced by individuals with vocal impairments, and our own experiences feeling that our speech was not heard clearly in our environment. So we decided to create a speech-to-text web app that facilitates communication between people with speech disabilities (regardless of whether its a stutter or speech from an autistic person).
What it does
The web app records words and gives a textual representation of the speech. Since we want to accommodate for a wide range in speech, the web app gives suggested text responses. In addition, if a word doesn't match any text responses that exist, a user can add the word into the system.
How we built it
Challenges we ran into
Our intention was to use Google's Speech Recognition API to process user audio recordings. We ran into issues using recorder.js, as the library gave a BLOB representation of recorded audio, but API calls to Google's API required a file in standard audio formats (eg. WAV, FLAC, etc). We were also limited by time constraints, as we couldn't build a full database that complied with our specs.
Accomplishments that we're proud of
- Learning to build a web app!
- Learning to debug
- Creating an amazing custom pizza at Blaze Pizza
- Sleeping :)
What we learned
- Developing is 90% debugging, 10% coding
- Things about life and cs in general
What's next for Vochnica
- Fully incorporating the Google Speech Recognition API