One thing discord fails at accomplishing is voice activation in their web browser client. If anyone from the company is reading I will gladly write you a python/scipy program that will do the correct real-time DSP for voice activation if you implement it into your platform.
It's a window, an fft, a filter around a primary frequency range, an integration, and an N second timer since the last time the audio crosses the threshold.
Discord's current implementation drops in the middle for words! It's crazy.
It's a window, an fft, a filter around a primary frequency range, an integration, and an N second timer since the last time the audio crosses the threshold.
Discord's current implementation drops in the middle for words! It's crazy.