3 ms·
This is some revisionist history right here. Speech recognition has gotten dramatically better since 1995. It now works with cheap mics, without per-speaker tr
by Eridrus 8y ago
This is some revisionist history right here.
Speech recognition has gotten dramatically better since 1995. It now works with cheap mics, without per-speaker training, across the room, in noisy environments, with large vocabularies and low error rates.
The problem with this article is that all of this tech is not a binary yes/no feature, but rather work with varying levels of success in various situations.
It also says ridiculous things like "Unlike today’s conversational interfaces, the Gemini took voice commands in an english-like programming language called VOCOL." as if that is a good thing, compared to these interfaces being usable in ambiguous natural language.
- tootie 8y agoI was using an early version of Dragon by Nuance which is still active. It used a cheap commodity microphone and hardware that, while expensive at the time, was maybe 10% as powerful as a typical modern phone. And having built a public-space voice system very recently, I can tell you that it's still really difficult to pull off and definitely requires specialty AV equipment and professional installation. The problem then as now is that if it's not 100% accurate, you still need to be doing manual correction which is very time-consuming. And typing was never really a broken system that needed help.