Hi john,

>I guess we'll have to wait and see.

>FWIW: for a long time, professionals had secretaries who used shorthand to record voice and convert it into typed documents. It worked well, but required highly skilled secretaries. More recently, portable recorder technology allowed professionals to do the same for remote transcriptionists. Shorthand vanished almost completely with transcription increasingly commoditized into typing pools. The digital age allowed those transcriptionists to be "lowest cost" providers in the third world.

>I think it is hubris to believe that technology cannot replicate the task of converting speech into text with appropriate punctuation and emphasis.

A lot of people have problems in speaking proper english with the proper punctuation. I know I still have problem in constructing proper english sentences and doing the proper spelling. And this fact applies to a lot of people. Certainly the software has to be aware of punctuation, culture, language and the jargon. working just like you in healthcare, I know a lot of american doctors comming from other countries like china, india, pakistan, etc. Even the kiwi lab director I know is difficult to understand for other people. You can understand what he is saying by listening by context and puzzling the words. And even then you could be wrong. And here I'm talking about someone who does speak native english (Actually that is worth a discussion :)).

I have no doubt that it eventually can be solved. I've evaluated a lot of speech to text implementations (for implementation of our software), but it really sucks. Most often they could not translate into the right words, the required training was not sufficent, and only if you speak very clear and slow you might get some words right, inmediately taking away the advantage. I'm not even talking making corrections here. It currently does not cut it. I don't think it will be solved easily. It requires a lot more than the current technology.

That same kiwi (i hope you're not insulted :) ) lab director has a wife who comes from down under. They now live in Toronto. In australia she worked for a private investigator who dictate his notes into a dictaphone and she processes the instructions (thus not only text) on there. She still does work for the guy and when he send off his WAV files to her at the end of the day, she will be just start working (because of the time difference) and she will have it finnished when he steps into the office the next day.

>Most of us can speak more quickly than we type and in highly resourced professions like healthcare, physicians prefer dictation wherever it is available. Make it more available and the preference will spread like wildfire. At which point much of the IT stuff that gets people all worked up today, will be like argument over which version of shorthand is the best. IMHO ;-)

That were nice, if 1. the software is capable enough and 2. if all its function is to write letters, not having assignments to the listener. Often the spoken text has to be made up in a certain document with a specific layout and some other addtional stuff as well. The doctor does not want to be bothered by that. He will ask the secretary to do this anyways. So the benefit of speech to text becomes really marginalised anyways.

Walter,