Terry Blanton <hohlr...@gmail.com> wrote:
> BTW, do you use Dragon Speak still? > Not that often. It is good for writing long, formal documents. Not so much for short messages. It is not good for editing papers, which is mostly what I do these days. I expect there will soon be AI versions of voice input transcription. Maybe online? I expect they will blow Dragon Speak out of the water. They will be far better. I say that for two reasons: ChatGPT translation from Japanese into English is far better than Google translate (https://translate.google.com). The Bots have generated a lot of accurate linguistic data. I expect it is nothing like human knowledge of grammar, but it works. I have recently discovered that AI based online OCR programs are far superior to desktop OCR programs such as Adobe Acrobat. I used Adobe Acrobat OCR to make old documents such as ICCF3 "searchable." You can export the resulting text to Microsoft Word or a text file. Recently ChatGPT recommended I try the Amazon AWS Textract online OCR program. I tried it. I did a file-compare of the AWS output compared to the Adobe Acrobat output. AWS has many fewer OCR errors. I think you could correct many voice input errors by using the AI linguistics-based methods, and the pattern recognition algorithms. I think the pattern recognition algorithms can be applied to audio track data in a way that is similar to images of printed letters.
