TannerPlauche / SpeechMapper

An experiment in speech analysis in the browser using the Web Audio API.

This is an experiment combining the web audio api and machine learning / artificial intelligence in the browser

Currently a hot mess of code. This is version is part of the learning process.

No transpiling needed.

There is currently no stored model. If you train for all current vowel training data, it takes at least 30 minutes.

If you train for minimal pairs (two vowels), only 5-10 minutes needed to train model.

Makes use of:

• JavaScript Digitial Signal Processing
• Brain.js for neural network machine learning