Work
Since it’s pretty cool-looking, I thought I’d share a screenshot of what I’ve been doing at work today:
The system (Steve’s) differentiates speech from background noise. This is a trivial test in a quietish room with one speaker (me) saying the same thing over and over again.
The top purple line is the system’s guess at where the speech is, with the bottom purple line being the real location of speech. The red and blue lines represent “likelihood that it’s speech” and “likelihood that it’s silence” – when the blue line is higher than the red for a significant time, the area is marked as speech.