What is this? What is this?!! My text data in the computer was never supposed to get this big. Megabytes. Okay, fine. I guess I had it coming for me if I know there are other people who have megabytes of text data inside the computer.
And, I guess in another sense, let that be a lesson to me. Let that be a lesson to me in second thinking that all the text data I would ever need, or even more so, that I would ever type out of my own fingers, would be smaller than some certain size. Because, as you can clearly see from this particular example, that is clearly not the case. And, it still doesn’t take into consideration, the fact that speech comes at a much higher rate and at a much greater frequency than text data. So this is only the tip of the iceberg in the severity of data. With an avalanche more of data to come once I turn on speech-to-text technologies.
So how do you do speech to text locally?
20160702/DuckDuckGo raspberry pi speech to text recorder
20160702/DuckDuckGo speech to text linux local train
20160702/http://www.daveconroy.com/turn-raspberry-pi-translator-speech-recognition-playback-60-languages/
20160702/http://diyhacking.com/best-voice-recognition-software-for-raspberry-pi/
20160702/https://en.wikipedia.org/wiki/List_of_speech_recognition_software
20160702/https://en.wikipedia.org/wiki/Julius_%28software%29
20160702/https://en.wikipedia.org/wiki/VoxForge
20160702/https://en.wikipedia.org/wiki/LibriVox
20160702/https://en.wikipedia.org/wiki/Virtual_volunteering
20160702/https://en.wikipedia.org/wiki/Kaldi_%28software%29
Search for “Asia_Online” above to find my related notes for running a local language translation service. Wow, so you’re saying you can run speech-to-text, language translation, and text-to-speech all locally? Yes, you can, but do you want to? It’s definitely not practical for the small amount of memory and computation that most mobile devices have. That’s why, as it turns out, that for mobile devices that claim to provide these services, many could just as well be running the actual translation within the cloud, and all the device does is capture and upload the data.