Kwongorera kuvuga
Kwongorera ni __AIU_PROTECTED_10 __ sisitemu ifunguye-sisitemu yo kumenyekanisha imvugo yikora ihindura amajwi mumyandiko mundimi 90+.
Incamake
It matters because it brought near-human transcription quality to everyone for free, working robustly on accents, background noise, and technical jargon.
Kwibira cyane
Yarekuwe na OpenAI muri Nzeri 2022, Whisper ni moderi ya Transoder ishingiye kuri encoder-decoder yatojwe kumasaha 680.000 yindimi nyinshi, amajwi menshi yakuwe kurubuga. Bitandukanye na sisitemu zabanje zikeneye amakuru asukuye, yanditseho, Whisper yigiye kubintu byafashwe amajwi ku isi, bituma idashobora kwihanganira imvugo, urusaku, n'inzira nyabagendwa. Icyitegererezo kimwe gikora transcript, guhindurwa mucyongereza, kumenyekanisha ururimi, no kugihe. Yohereza mubunini kuva kuri 'gito' (ibipimo 39M) kugeza kuri 'binini' (1.55B), bigatuma abakoresha ubucuruzi bwihuta kubwukuri. Kuberako ibipimo byemewe kumugaragaro munsi ya MIT, Whisper yabaye umugongo wambere kubatwara podcast zitabarika, ibikoresho byanditse, hamwe na porogaramu zijwi hafi ijoro ryose.
Ubushishozi
Whisper igabanya amajwi mo ibice 30-isegonda, ihindura buriwese muri log-Mel spectrogramu (imiyoboro 80 yumurongo), ikayigaburira kodegisi ya Transformer. Decoder noneho ihanura ibimenyetso byanditse byikora, iyobowe nibimenyetso byihariye byerekana inshingano (transcription vs. translate), ururimi, ndetse no kohereza igihe cyagenwe. Iyi multitask token-conditioning nuburyo bwubwenge: umurongo umwe wibiro ukora imirimo myinshi bitewe nibimenyetso byihuse byatanzwe mugitangira decoding.
Ingaruka z'Ingamba
Kugera no kugera
Itezimbere kugerwaho binyuze mu kwandukura, kuvuga, no guhuza amajwi.
Igiciro na bije
Amatsinda yibitangazamakuru arashobora kohereza amajwi yihuse hamwe na bije nto.
Umuvuduko n'igipimo
Sisitemu ireba abakiriya irashobora gutunganya imikoranire ivugwa murwego runini.
Igihe kizaza cyo Kwongorera Imvugo
Whisper yateje umurongo wibikomoka byihuse nka Whisper.cpp, byihuta-byongorerana, hamwe na verisiyo itandukanye ikora mugihe nyacyo kuri terefone na mudasobwa zigendanwa. Tegereza uburyo bworoshye (buke-bwihuta) butandukanye, disikuru nziza yo kuvuga ihujwe nayo, hamwe nibikorwa bikomeye kururimi ruke. Mugihe amajwi-y-amajwi AI akura, moderi yoroheje ya Whisper-moderi irashobora gukoresha imbaraga zanditseho, inyandiko zinama, hamwe nibikoresho byifashishwa kuri interineti rwose, bikarinda ubuzima bwite mugihe bihuye nukuri kurwego rwibicu.
Gushyira mu bikorwa Isi
Kwikora-gukora-gushakisha inyandiko-mvugo hamwe nibisobanuro bya podcasts na videwo ya YouTube
Guha imbaraga imbonankubone-inoti porogaramu zitanga incamake kuva Zoom cyangwa Amakipe amajwi
Guhindura ibiganiro byindimi zamahanga mu nyandiko yicyongereza kubanyamakuru
Kubaka amajwi agenzurwa nibikoresho byifashishwa hamwe nigitekerezo kubakoresha badashobora kwandika
Ingaruka & Kurinda
Gukoresha nabi amajwi no kwigira ibyago byiyongera mugihe uruhushya rubuze.
Ukuri kurashobora kugabanuka hejuru yimvugo, imvugo, cyangwa urusaku rwibidukikije.
Amajwi yubukorikori arashobora kwibeshya kumvugo yukuri nta kirango gisobanutse.
Igishushanyo mbonera
Shaka uruhushya rusobanutse rwo gufata amajwi, gukoroniza, no gukoresha.
Ikizamini cyiza mubiganiro bitandukanye hamwe nuburyo bwimbere.
Sobanura igihe umuntu agomba gusuzuma cyangwa kwemeza ibisubizo.
Andika amajwi yubukorikori kandi ugumane inyandiko zerekana kubazwa.
Komeza Ubushakashatsi
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 Whisper Speech Recognition 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
Ubuyobozi bukurikira
Kumenyekanisha Amarangamutima
Ibibazo bikunze kubazwa
What is Whisper Speech Recognition?
Kwongorera ni __AIU_PROTECTED_10 __ sisitemu ifunguye-sisitemu yo kumenyekanisha imvugo yikora ihindura amajwi mumyandiko mundimi 90+. Nibyingenzi kuko byazanye ubuziranenge bwabantu-bantu kubantu bose kubuntu, gukora cyane kumurongo, urusaku rwimbere, na jargon ya tekiniki.
Hafi y'amasaha angahe y'amajwi Whisper yatojwe?
Whisper yatojwe kumasaha agera kuri 680.000 yindimi nyinshi, amajwi menshi yakusanyirijwe kurubuga, imibare nini idasanzwe kandi itandukanye.
Ni ubuhe buryo bwo kugereranya Whisper ibara uhereye kumajwi mbisi mbere ya kodegisi?
Whisper ihindura buri gice cyamajwi 30-amasegonda mumirongo 80 ya log-Mel yerekanwe, iyo kodegisi ya Transformer itunganya.
Nigute moderi imwe ya Whisper ikora imirimo myinshi nko kwandukura no guhindura?
Kwongorera ibintu ku kimenyetso kidasanzwe mugitangira decoding ibwira ururimi, umurimo, ndetse no kohereza igihe.
Ni ubuhe buryo rusange bwubaka imitsi Whisper ikoresha?
Whisper ni encoder-decoder Transformer: encoder isoma spekrogramu kandi decoder itanga ibimenyetso byanditse byikora.
Kuki Whisper ifatwa nkigikomeye cyane ugereranije na sisitemu ya mbere ya ASR?
Amahugurwa kumubare munini wamajwi atandukanye, nyayo-yisi (amajwi, urusaku, umuhanda) byahaye Whisper imbaraga zikomeye ziva mumasanduku idafite imbaraga kuri buri murongo.