History Windows Speech Recognition

1 history

1.1 precursors
1.2 development
1.3 security report

history
precursors

microsoft has been involved in speech recognition , speech synthesis research many years. in 1993, microsoft hired xuedong huang carnegie mellon university lead speech development efforts. company s research led development of speech api, introduced in 1994. speech recognition technology has been used in of microsoft s products prior windows speech recognition. versions of microsoft office, including office xp , office 2003, included support speech recognition among office applications , other applications such internet explorer. installation of office enable limited speech functionality in windows nt 4.0, windows 98 , windows me. 2002 edition of windows xp tablet pc edition include support within tablet pc input panel feature, , microsoft plus! windows xp expansion package enabled voice commands used in windows media player. however, support limited individual applications, , prior windows vista, windows operating system did not include integrated support speech recognition.

development

a prototype speech recognition aero wizard interface seen in windows vista (then known longhorn ) build 4093.

at windows hardware engineering conference of 2002, microsoft announced windows vista, known codename longhorn, include advances in speech recognition technology , features such support microphone arrays. bill gates expanded upon information during professional developers conference of 2003 stated company build speech capabilities system -- big advance in longhorn, in both recognition , synthesis, real-time. further reports said operating system include integrated support speech recognition, , pre-release builds throughout development of operating system include speech engine training features. in 2003, microsoft clarified extent of intended integration windows vista when company stated within pre-release software development kit common speech scenarios, speech-enabling menus , buttons, enabled system-wide in operating system.

during winhec 2004, microsoft listed speech recognition part of longhorn mobile pc strategy improve productivity , listed microphone arrays hardware opportunity operating system. @ winhec 2005, microsoft released additional details pertaining speech recognition in windows vista focus on accessibility, new mobility scenarios, , improvements speech user experience. unlike speech support included in windows xp, integrated tablet pc input panel , required switching between dictation , command modes, windows vista separate feature tablet pc input panel introducing dedicated interface speech input on desktop , unify separate dictation , command modes. in previous versions of windows, speech recognition not allow user speak command after dictation or vice versa without first switching between these 2 modes. microsoft stated speech recognition in windows vista improve dictation accuracy, , support additional languages , microphone arrays. demonstration of feature @ winhec 2005 focused on e-mail dictation correction , editing commands, , set of slides dedicated microphone arrays released. windows vista beta 1 include integrated speech recognition application. in effort persuade company employees interact windows speech recognition during development, microsoft offered opportunity win premium model of xbox 360 video game console.

on july 27, 2006, before operating system s release manufacturing (rtm), notable incident pertaining speech recognition occurred during demonstration microsoft @ annual financial analyst meeting. speech recognition failed function correctly, resulted in unintended output of: dear aunt, let s set double killer delete select when several attempts dictate led consecutive output errors; incident subject of significant derision among analysts , journalists present in audience. microsoft later revealed errors during demonstration due audio gain glitch caused speech recognition distort dictated commands. glitch fixed prior operating system s release manufacturing on november 8, 2006.

security report

reports surfaced in 2007 windows speech recognition may vulnerable attack allow attackers take advantage of capabilities perform undesired operations on targeted computer playing audio through targeted computer s speakers; first vulnerability discovered after operating system s general retail availability. while microsoft stated such attack theoretically possible, have meet number of prerequisites in order successful: targeted system required have speech recognition feature activated , configured, speakers , microphone(s) connected targeted system need turned on, , exploit require software interpret commands without user noticing—an unlikely scenario affected system perform user interface operations , produce audible feedback (as speakers need active). moreover, mitigating factors include dictation clarity, , microphone feedback , placement. exploit of nature not able perform privileged operations users or protected administrators without explicit user consent because of user account control.