20 years ago you spoke directly into a microphone and could only use an extremely limited set of supported languages and locales/accents. You're also missing t

20 years ago you spoke directly into a microphone and could only use an extremely limited set of supported languages and locales/accents. You're also missing the point. I don't think anyone has ever claimed that speech recognition can only be done on supercomputers. Your laptop can surely run one of these models (though it would take a long time to train one). But there's a reason why an Echo Dot cost $20 and not $1000.