Natural Language and Speech parts of Chapters 8

  • Slides: 14
Download presentation
Natural Language and Speech (parts of Chapters 8 & 9)

Natural Language and Speech (parts of Chapters 8 & 9)

Natural Language in Computing • Use of human languages – Translation – Commands to

Natural Language in Computing • Use of human languages – Translation – Commands to computer – Queries – Text Database Searching – Text generation – Games

Commands to Computer • From computer-oriented to domain oriented • May not be more

Commands to Computer • From computer-oriented to domain oriented • May not be more efficient than selection • Speech recognition may help

Natural Language Queries • Limited form of commands to computer • Actions able to

Natural Language Queries • Limited form of commands to computer • Actions able to be requested are database queries (searches) • Experienced users shorthand • Things aren’t as grim as Shneiderman makes it seem

Text Database Searching • DB contains text as main content • Common goal is

Text Database Searching • DB contains text as main content • Common goal is retrieval of relevant records using natural language question • Meaning vs matching • Statistical • Pre-processing • Information retrieval contests • Information Push, E-mail filtering

Natural Language Text Generation • Output in text, frequently from data • Generation of

Natural Language Text Generation • Output in text, frequently from data • Generation of poems and stories • Conversational systems

Games • Command based games

Games • Command based games

Speech Recognition, Digitization and Generation • Speech recognition progress is slow • Challenges –

Speech Recognition, Digitization and Generation • Speech recognition progress is slow • Challenges – background noise, speaker variation • Drawbacks – human memory use • Benefits – accommodation of disabilities, environment/task requirements • Growth – now many products

Discrete Word Recognition • Sentences spoken is slow deliberate manner – with words being

Discrete Word Recognition • Sentences spoken is slow deliberate manner – with words being discrete entities, rather than run together (continuous) • Not tolerable for most people • Discrete word recognition is easy • Continuous speech recognition is harder

Speech Store and Forward • Store and forward spoken messages • Could be used

Speech Store and Forward • Store and forward spoken messages • Could be used for groupware – computer supported cooperative work

Speech Generation • • Very feasible – done all the time Can be annoying

Speech Generation • • Very feasible – done all the time Can be annoying / noisy Valuable for handicapped Completely computer generated vs human sounds pieced together vs stored words

When to Use Speech • Message is simple • Message is short • Message

When to Use Speech • Message is simple • Message is short • Message will not be referred to later • Message deals with events in time • Message requires an immediate response • Visual channels of communication are overloaded • Environment is unsuitable for transmission of visual info • User must be free to move around

Audio Tones and Music

Audio Tones and Music

End Speech and Natural Language

End Speech and Natural Language