3 ms·
Why is the cloud required for speech to text when a four core ARM SOC is under 15 dollars? My Commodore 64 had good text to speech, and Dragon was doing speech
by api 12y ago
Why is the cloud required for speech to text when a four core ARM SOC is under 15 dollars? My Commodore 64 had good text to speech, and Dragon was doing speech to text on 90s PCs. I don't get the technical rationale.
- DanBC 12y agoYou needed to train Dragon and you needed to calibrate the microphone. People talking to control their tv's want to be able to iist talk. Thus, instead of training the software you offload that training to the cloud and massive computing to do it. I agree that the tv setup could include a bit of voice recognition training. But then the TV only changes channels if Ann asks it to. Bob's out of luck, he has to use the remote.
- api 12y agoAhh... that explains it a little more... though I don't see why you couldn't just share model data via the cloud instead of actually sending audio from a microphone directly out to a remote endpoint. But then again anything with an Internet connection and a mic (laptop, cell phone, etc.) is a potential spy device with the right malware installed.