Keyword Spotting and Wake Words
Ukubona igama elingukhiye ubuchwepheshe obuhlala bulalela obuvumela idivayisi ukuthi ilinde umusho owodwa wokucupha njengokuthi 'Hey Siri' noma 'Alexa' ngaphambi kokuqalisa esenzweni.
Uhlolojikelele
It matters because it makes hands-free voice control possible while keeping power use and privacy intrusion low.
I-Deep Dive
I-wake-word detector imodeli encane, ekhethekile yenkulumo okuwukuphela kwayo umsebenzi wokuphendula umbuzo owodwa izikhathi eziningi ngomzuzwana: ingabe umsebenzisi uvele asho umusho wokucupha? Ngokungafani nokubonwa kwenkulumo okugcwele, ayilobi yonke into — isebenzisa inethiwekhi encane ye-neural ngqo kudivayisi, iskena amawindi amafushane agqagqene omsindo. Ukuze wonge ibhethri, amafoni nezipikha ezihlakaniphile ngokuvamile zisebenzisa idizayini enezigaba ezimbili: i-chip yamandla aphansi kakhulu ilalela okufanayo, bese ivusa imodeli ethe ukukhudlwana ukuze iqinisekise ngaphambi kokusakaza noma yini emafini. Onjiniyela bashuna umkhawulo ukuze balinganisele ukwamukela okungamanga (ukuvuka uma kungabizwa muntu) ngokumelene nokulahlwa okungamanga (ukuziba umyalo wangempela), futhi baqeqeshelwa izinkulungwane zamaphimisela, amabanga, namagumbi anomsindo.
I-Technical Insight
Umsindo ongenayo usikwa ube ngu-~20-40 millisecond frames futhi uguqulelwe izici ezifana nama-MFCC noma amandla e-mel filterbank. Inethiwekhi ye-neural ehlangene — ngokuvamile imodeli encane eguquguqukayo noma ephindaphindayo, ngezinye izikhathi isebenzisa ama-convolutions ahlukaniseka ngokujulile ukuze inciphe usayizi — ikhipha amathuba ebinzana eliqondiwe kuhlaka ngalunye. Isinyathelo sewindi elishelelayo noma elislayidayo livimbela amafreyimu angawodwa anomsindo ukuthi angabangeli, futhi ukubona kubasa kuphela lapho ukuzethemba kuhlala kuphezulu kumafreyimu alandelanayo.
I-Strategic Impact
Finyelela futhi ufinyelele
Ithuthukisa ukufinyeleleka ngokuloba, ukulandisa, nezixhumi ezibonakalayo zezwi.
Izindleko kanye nesabelomali
Amaqembu emidiya angathumela umsindo opholishiwe ngokushesha ngamabhajethi amancane.
Isivinini nesikali
Amasistimu abhekene nekhasimende angacubungula ukusebenzelana okukhulunyiwe ngesilinganiso esikhulu.
Ikusasa Lokubheka Amagama Angukhiye kanye Namagama Okuvusa
Amamodeli e-wake-word aya eba mancane futhi awabantu. Ukufunda kudivayisi kuzokuvumela ukuthi ubhalise imishwana yokuqalisa ngokwezifiso futhi ujwayelane nezwi lakho ngaphandle kokuthumela umsindo noma kuphi. Lindela ukuhlanganiswa okuqinile nge-silicon enamandla aphansi 'ehlala ivuliwe', izibangeli zokushintsha izilimi eziningi kanye nekhodi, nokuqina okungcono kuma-TV, umculo, nomsindo ovela kude. Imiklamo egcina ubumfihlo egcina konke ukulalela kwendawo - eqinisekisa izwi lokuvuka ngaphambi kwanoma yimuphi othintana naye wenethiwekhi - isiba okulindelekile okuzenzakalelayo.
Ukuqaliswa Komhlaba Wangempela
Ukuthi 'Alexa' ku-Amazon Echo noma 'Hey Google' kusipika se-Nest ukuze uqale isicelo sezwi se-handsfree
'Hey Siri' evusa i-iPhone noma ama-AirPods endaweni ekhiyiwe, enamandla aphansi ngaphandle kokucindezela inkinobho
Amasistimu we-infotainment yemoto alalele umusho othi 'Hey Mercedes' ukuze abashayeli bakwazi ukulungisa ukuzulazula ngaphandle kokususa izandla esondo
Amahedisethi esibhedlela nawesitolo asebenza emyalweni okhulunywayo ukuze abasebenzi bakwazi ukungena idatha begqoke amagilavu futhi izandla zigcwele
Izingozi & Guardrails
Ukusetshenziswa kabi kwezwi kanye nezingozi zokuzenza ongeyena ziyanda uma imvume ingekho.
Ukunemba kungase kwehle kuzo zonke izinhlobo zokuphimisela, izilimi zesigodi, noma izindawo ezinomsindo.
Umsindo wokwenziwa ungenziwa iphutha njengenkulumo eyiqiniso ngaphandle kokulebula okucacile.
Ukuqalisa Umhlahlandlela
Thola imvume esobala yokuthwebula izwi, ukuhlanganisa, nokusebenzisa kabusha.
Ikhwalithi yokuhlola kuzo zonke izipikha nezimo zangemuva.
Chaza ukuthi kunini lapho umuntu kufanele abuyekeze noma agunyaze okuphumayo.
Lebula umsindo wokwenziwa futhi ugcine amarekhodi atholakalayo ukuze aziphendulele.
Qhubeka Uhlole
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Keyword Spotting and Wake Words quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Umhlahlandlela olandelayo
Ukuqondanisa Kwegama Elinesitembu Sesikhathi
Imibuzo evame ukubuzwa
What is Keyword Spotting and Wake Words?
Ukubona igama elingukhiye ubuchwepheshe obuhlala bulalela obuvumela idivayisi ukuthi ilinde umusho owodwa wokucupha njengokuthi 'Hey Siri' noma 'Alexa' ngaphambi kokuqalisa esenzweni. Ibalulekile ngoba yenza ukulawula kwezwi kwe-hands-free kwenzeke ngenkathi igcina ukusetshenziswa kwamandla nokungena kobumfihlo kuphansi.
Uyini umsebenzi oyinhloko womtshina we-wake-word?
Umtshina we-wake-word awulobi konke; ikhombisa kuphela uma kukhulunywa ibinzana lokucupha elikhethiwe ukuze isistimu enkulu ikwazi ukuvuka.
Kungani izikhulumi eziningi ezihlakaniphile zisebenzisa idizayini yokuthola izigaba ezimbili?
Isiteji esincane samandla aphansi siyalalela njalo futhi sivusa imodeli enekhono kakhulu ukuze siqinisekise, okugcina ukusetshenziswa kwamandla kuphansi kakhulu.
Kusho ukuthini 'ukwamukela okungamanga' ekutholweni kwegama lokuvuka?
Ukwamukela okungamanga kuyimbangeli engadingeki - isistimu iyasebenza lapho igama lokuvuka lingashiwongo. Okuphambene ukwenqatshwa okungamanga.
Cishe ulungiswa kanjani umsindo ongenayo ngaphambi kokuthi inethiwekhi ye-neural iwubone?
Umsindo uhlukaniswa ube ozimele abafushane (amashumi ama-millisecond) futhi uguqulwa ube izici ezihlangene imodeli engakwazi ukwenza amaphuzu ngohlaka ngohlaka.
Kungani ukutholwa ngokuvamile kudinga ukuzethemba okuphezulu kozimele abambalwa abalandelanayo?
Ukushelela kumafreyimu amaningi kumisa ukukhuphuka komsindo okufushane ekudubuleni umtshina, okuthuthukisa ukwethembeka.