Voice Lab

Hum, sing, or beatbox and have it turned into digital sounds that can be accessed and manipulated with the Web Audio API. Great for game developers or other creators who aren't musically inclined.

Comment

Inspiration

In my last codex project (for the reddit games with a hook hackathon) I used the Web Audio API to generate sound effects and music. I had a hard time picturing what things would sound like just by editing the config being sent to the API, so I had codex build a lightweight website so I could use sliders to adjust all of the different parameters and get the sound I wanted. When I discovered this hackathon, I thought about that project and decided it would be cool if I could hum a melody or create a beat with my voice and have that transcribed into something digital that the Web Audio API could accept.

What it does

It records a short clip of your voice humming a melody, making a sound effect, or creating a beat, sends it to a few different apis for analysis, and then generates a config for the Web Audio API to attempt to recreate the input. You can edit the config, and add the results of multiple configs to a primary track.

How we built it

I used Codex and VS Studio Code.

Challenges we ran into

The analysis of the audio files was tricky and I had to try a lot of different APIs, and honestly the results still aren't where I want them to be. Music and audio files are trickier and more complex than I realized they would be.

Accomplishments that we're proud of

Even though the output isn't as good as I'd like, the actual UI/UX and functional flow I think is intuitive and powerful.

What we learned

I learned a bit about some of the different audio analysis APIs that are available, and how to pick apart sound effects/beats/melodies.

What's next for Audio Lab

I will likely look for an audio engineer I can work with who will have some insights into any limitations of my current workflow.

Built With

Share this project:

Updates