Speech recognition is terrible on specialised vocabulary, strong accents, low quality audio, etc. You can train it, but only if you have transcriptions of your audio. Steno is a good way to make transcriptions.

I have mathematics education videos I would like subtitles for, all in non-US, non-UK accents, full of specialised terminology. In my tests, Amazon and Google's APIs were completely useless for this, as was Dragon. If I find the time, I would like to learn steno of some sort to make the subtitles and provide training data for future videos.

Amazon’s API fails any time someone uses a slightly less than super common name. It was really depressing.

Is Otter.ai's tech any better?

It will be relevant, because there will be people interested in typing. It is like how there are pencil artists even though we can all use Photoshop.

For myself: how I think when I’m writing is somewhat different than when I’m speaking. I still prefer writing _and_ typing despite the ability to record myself.

I also think we’re a bit of a ways off before speech-to-text will work for programming…

Maybe what I’m looking for isn’t stenography, but chorded typing with the ability to break out of it when I need to.

The technology is nowhere near a human level yet. There are people who use it, but in order to aid recognition they clearly repeat everything being said into a "steno-mask" for software transcription: https://en.wikipedia.org/wiki/Stenomask

Still will be a long, long time for such things to be viable in a courtroom.