UMHLAHLANDLELA WE-AI womsindo

Ukuhlukaniswa Kwenkulumo kanye Nenkinga Ye-Cocktail Party

Ukuhlukaniswa kwenkulumo kuwumsebenzi wokudonsa amazwi ngamanye ngaphandle kokurekhodiwe lapho abantu abambalwa bekhuluma ngesikhathi esisodwa.

2 amaminithi ukufundaIgcine ukubuyekezwa

Uhlolojikelele

It tackles the 'cocktail party problem' that humans solve effortlessly but machines find genuinely hard.

I-Deep Dive

Ephathini enomsindo, ungagxila engxoxweni eyodwa kuyilapho ukuhlunga okunye, ikhono isazi sokusebenza kwengqondo u-Colin Cherry saqamba ngokuthi 'inkinga yephathi ye-cocktail' ngo-1953. Amakhompiyutha athwala kanzima ngenxa yokuthi amazwi agqagqene ahlangana abe i-waveform eyodwa, futhi uhlelo alazi kusengaphambili ukuthi zingaki izikhulumi ezikhona noma ukuthi imuphi umsindo okabani. Ama-algorithms okuhlukanisa inkulumo athatha lowo msindo oxubile futhi akhiphe ithrekhi ehlukile, ehlanzekile yesipika ngasinye. Izindlela zakuqala zazisebenzisa izindlela zezibalo kanye nezinhlu zamakrofoni ukuze zisebenzise izinkomba zendawo. Ukuphumelela kufike namamodeli okufunda ajulile afana ne-Deep Clustering kanye ne-TasNet/Conv-TasNet, afunda ukufihla noma ukwakha kabusha izwi ngalinye ngokuqondile ku-waveform, ngisho nombhobho owodwa.

I-Technical Insight

Amasistimu amaningi asebenza esizindeni esifundiwe noma se-spectrogram: inethiwekhi ye-neural ilinganisela 'imaski' yesipikha ngasinye okuthi, uma sisetshenziswa kungxube, ihlukanise lelo zwi. Amamodeli esizinda sesikhathi afana ne-Conv-TasNet eqa i-spectrogram ngokuphelele futhi asebenze kumasampuli aluhlaza ukuze uthole ukwethembeka okuphezulu nokubambezeleka okuphansi. Inselele eyinhloko inkinga yokuvumela, enquma ukuthi imaphi amamephu wesiteshi okukhiphayo ukuthi isiphi isipikha, esixazululwa ngokuqeqeshwa okungaguquguquki kwemvume ukuze imodeli ingajeziswa ngoku-oda okukhiphayo.

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 Lokuhlukaniswa Kwenkulumo kanye Nenkinga Yephathi Ye-Cocktail

Ukwehlukana kuya ezimeni ezivulekile, zomhlaba wangempela: izinombolo ezingaziwa nezishintshayo zezikhulumi, amagumbi anomsindo, nomsindo osakazwayo oqhubekayo. Isizinda sesipikha esiqondiwe, lapho unikeza khona imodeli isampula yezwi elifushane ukuze ikhiphe lowo muntu, ikhuphuka ngokushesha. Amamodeli ahlanganisiwe omsindo nokubukwayo asebenzisa ukunyakaza kwezindebe ukuze ahlukanise amazwi. Lindela lawa makhono ashumekwe kuzinsiza-kuzwa, ama-earbud, nokulotshwa komhlangano, okuvumela amadivayisi agqamise noma ubani ofuna ukuzwa.

Ukuqaliswa Komhlaba Wangempela

Amathuluzi okuloba omhlangano ahlukanisa izipikha ezeqayo ukuze amagama omuntu ngamunye abalulwe ngendlela efanele kumanothi.

Izinsiza-kuzwa ezithuthukisiwe zihlukanisa isikhulumi esisodwa endaweni yokudlela egcwele abantu ukuze yenze ingxoxo ibe lula kumuntu ozigqokile.

Ukukhiqizwa komculo ne-podcast kusebenzisa ukuhlukanisa ukuhlukanisa amazwi ezinsimbini noma ukuqaqa ukungqubuzana phakathi kwabasingathi.

Amapayipi okubona inkulumo ahlukanisa kuqala umsindo oxubile ukuze izwi ngalinye lilotshwe ngokunembile.

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

1

Thola imvume esobala yokuthwebula izwi, ukuhlanganisa, nokusebenzisa kabusha.

2

Ikhwalithi yokuhlola kuzo zonke izipikha nezimo zangemuva.

3

Chaza ukuthi kunini lapho umuntu kufanele abuyekeze noma agunyaze okuphumayo.

4

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 Speech Separation and the Cocktail Party Problem quiz

Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.

Qala imibuzo

Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation

Umhlahlandlela olandelayo

Ukuhlukaniswa Komthombo Womculo weDemucs

Imibuzo evame ukubuzwa

What is Speech Separation and the Cocktail Party Problem?

Ukuhlukaniswa kwenkulumo kuwumsebenzi wokudonsa amazwi ngamanye ngaphandle kokurekhodiwe lapho abantu abambalwa bekhuluma ngesikhathi esisodwa. Ibhekana 'nenkinga yephathi ye-cocktail' abantu abayixazulula kalula kodwa imishini ikuthola kunzima ngempela.

Ithini 'inkinga yephathi ye-cocktail'?

Iqanjwe uColin Cherry ngo-1953, ichaza inselele yokulalela isikhulumi esisodwa lapho abantu abaningi bekhuluma ngesikhathi esisodwa.

Uyini umphumela wesistimu yokuhlukanisa inkulumo uma kubhekwa ukurekhodwa okuxubile kwezikhulumi ezimbalwa?

Ukwehlukana kukhiqiza ukusakaza okukodwa okuhlanzekile kusipika ngasinye, kuqaqa amazwi agqagqene kungxube.

I-Conv-TasNet yenzani ngokuhlukile ezindleleni eziningi zokuhlukanisa zangaphambilini?

I-Conv-TasNet isebenzisa isifaki khodi esifundiwe kumsindo ongahluziwe kune-spectrogram engaguquki, evumela ukwethembeka okuphezulu, ukuhlukaniswa kwe-latency ephansi.

Ingabe amanethiwekhi amaningi okuhlukanisa ahlukanisa kanjani izwi ngalinye exubeni?

Imodeli ibikezela imaski ngesipika ngasinye; ukukusebenzisa enhlanganisweni kugcina okuqukethwe yileso sikhulumi futhi kucindezela okunye.

Kuyini 'ukukhishwa kwesikhulumi esiqondiwe'?

Esikhundleni sokuhlukanisa wonke umuntu, unikeza inkomba yezwi futhi imodeli ikhipha leso sipikha esiqondiwe kuphela.