GOOD NEWS, BAD NEWS:

A QUICK SURVEY OF WINDOWS SPEECH RECOGNITION

JULY 2006, Updated November 2006, April 2007, June 2009

Itamar Even-Zohar

 

Good News

Bad News

Comments

New recognition engine (8.0).*

 

·       Recognition level in English is not remarkably more advanced than Microsoft Speech 2003 (which is however good). (The recognition level for the other languages is higher.)

·       Inflexible link between English variety and spelling.

·       No considerable amplification of vocabulary and databasing of names.*

 

* Engine in Office 2003 is 6.1

 

 

* MS maintains the parochial tradition to package a language variety with a particular spelling.  If you speak more like UK but prefer US spelling you are not served by this version, because each English module is now more variety-specific.

Many new and useful commands have been added, mostly for navigation and a better text handling. Some unique to MS.

No proper configuration options.

The new set of commands seems to be the best that Vista speech has to offer.

The separate dictation and command modes are now fused to one mode.

·       No ability to save speech data.

·       No playback.

·       No text-to-Speech Button.

All these missing features ARE operative in SR Office 2003!

 

Playback in Vista is limited to segments marked for correction in the Correction Dialogue.

A Correction Dialogue

·       No line for selection and modification by voice or typing in the Correction Dialogue.

·       No Correction Button

A Correction Button is operative in SR Office 2003

 

No Hot Keys. Only <CTRL-Windows> available as a toggle hot key for microphone. No other keys are customizable.

 

SR Office 2003 has F11 and F12 and allows customization.

 

The <Windows> key is not available on all keyboards. On some Laptops, however, it can be customized with a keyboard utility application (such as provided by IBM/Lenovo for its laptops).

 

A hot key is indispensable for situations where SR must be instantaneously stopped.

Works in many applications.

However, would not work in applications that do not currently support something Microsoft calls Text Services Framework. This is a standardized Windows way for applications to expose the text in their documents so that technologies like Speech Recognition and Handwriting Recognition can interact with those applications. (Rob Chamber’s explanation.)

 

 

No way to convert text to numbers or numbers to text by voice.

 

This feature is available in DNS.

A full macro language is available as an add-on. The current version supports all languages. To download it, click here. For a library of macros go to the Yahoo Group Files Section.

For a Manual and other documentation about macro writing go to the LINKS section of the Yahoo Group.

A Save/Restore add-on for saving and restoring User Profiles was made available in June 2009. To download it, click here

Profile Manager available as an add-on in SR Office 2003. Also available in DNS.

Supports 6 languages.

Does not provide quick and easy switching procedure between languages.

For instructions on handling languages in Vista Speech see my document “Multilingual Language Use in Vista Speech”.

 

No Transcription.

Feature fully functional in DNS.

Extensive Command-and-Control

 

Full mouse and keyboard emulation.

Meant to work in all applications and with all Windows screens. In practice still only partly functional.

Disabled people will be able to do quite a lot of C&C with it. Others would probably prefer partial use.

 

Documentation is still very scant. Even the list of available (and wonderful) commands is not complete.