Ceci est une ancienne révision du document !
[This article was dictated, not written. Mistakes are noted in brackets.]
Several months ago I was approached by one of the staff (last?) if I could find a solution for a participant who needed to dictate a novel. The participant didn't have a lot of resources so we needed to find a cost effective way of doing this. Of course when it comes to low-cost computers Linux is always a great choice(,) since we don't have to deal with licensing costs. We started by looking at commercial solutions that ran on Windows and even though there was some money available for these solutions there really wasn't money to test the solutions(,) so we needed something that we could try before we purchased it to make sure that it actually worked.
The problem with a lot of voice dictation solutions is that they're not very good at recognizing speech. We tried several solutions before we found one that [Sammy] semi-accurately recognize our speech. We also needed something that was very simple to use. Unfortunately some of the solutions were quite (L)imited in their use. One program [when] we used would only let us create a single document. This wasn't a great solution since the person wanted to create several stories in addition to their novel.
The solution we came up with was to use Google Docs in addition to the speech recognition add-on. To enable the add-on first go to the add-ons menu in Google Docs word application and click get add-ons. Next search for the speech recognition text by efv-solutions.com. Unfortunately this add-on won't work (with in) Firefox so you'll need either Google Chrome or the Chromium open source web browser.
To enable speech within a document click on the add-ons menu then click on speech recognition and click on the start Sub menu. Chrome or Chromium will ask you if you want to allow Chrome or Chromium to access the microphone - click yes. On the right hand side a window will appear where you can choose the language and dialect. In our case we selected English Canada. Then click the blue Start button on the left hand side of the screen and start dictating.
The speech recognition add-on has a limited amount of configurability. To configure the add-on, click Add-ons, Speech Recognition and Configure. This will bring up a pop-up window where you can configure the number of spaces after a period, the [work] word for period, the word for a new line, the word for a comma, the word for a semi-colon, and the word for a colon. While it’s a very limited configuration, it seems to be just enough to cover most of the basics.
So just how accurate is the speech recognition add-on well most of this document was created using the speech recognition at all I simply corrected a few things which you'll see by the brackets around different words. The most difficult thing was actually my grammar and dictating what was on my mind with a rambling on too much. The speech recognition plug-in has a fairly High rating in the Google app Store and there's a good reason for this it's actually a fairly decent plug-in despite its simplicity. It does the job it's not a great job I'm sure I can type faster and more accurately but it's accurate enough and surprisingly accurate enough that I decided I would actually use it for most of this document.
If you don't say anything for about 30 seconds this speech recognition turns itself off and you have to click start again. I found it was fairly good at detecting rapid speech as well as normal speech and slow speech. This add-on was created by [elizardo] Elisardo Felix and to date has no donations on [pleasure.com] pledgie.com but it's certainly deserves it given the level of accuracy.
Resources:
EFV Solutions: http://efv-solutions.com/