Voice Recognition for Humanity
Big kudos to Mozilla for launching a project to open source voice interaction. Called Project Common Voice, it will allow people to train and verify speech samples that can then be used for building open speech recognizers.
While open ASRs like Kaldi and Sphynx have been around for some time, the data to train them has not. With this project, it now means that it’s possible for an open source software to have comparable performance.
This brings up an interesting scenario… what if AI could be open and democratized like Wikipedia? What if it were massively distributed so that no one company or could wield it for its own advantage?
Common Voice is dangerous to Amazon, Google, and Apple. Why? It takes their data stream away. What if those who used Common Voice’s eventual APIs would be contributing to making a system for all humanity better? Best, it wouldn’t be for pushing us products or services.