Mmata Okwu ntanye
Whisper bụ OpenAI sistemụ njirimara okwu akpaka nke mepere emepe nke na-atụgharị ọdịyo ka ọ bụrụ ederede n'ofe asụsụ 90+.
Nchịkọta
It matters because it brought near-human transcription quality to everyone for free, working robustly on accents, background noise, and technical jargon.
Ime miri emi
Nke OpenAI tọhapụrụ na Septemba 2022, Whisper bụ ihe ngbanwe dabere na ihe ngbanwe a zụrụ azụ na awa 680,000 nke asụsụ ọtụtụ, ọdịyo multitask ewepụrụ na webụ. N'adịghị ka sistemu ndị mbụ chọrọ data dị ọcha, nke akpọrọ, Whisper mụtara site na ndekọ ndekọ ụwa na-adịghị mma, na-eme ka ọ na-agbanwe nke ukwuu maka ụda olu, mkpọtụ, na okwu mkparị. Otu ihe nlereanya na-ejikwa ndegharị, ntụgharị asụsụ n'ime Bekee, njirimara asụsụ, na nhazi oge. Ọ na-ebugharị na nha site na 'obere' (39M parameters) ruo 'nnukwu' (1.55B), na-ahapụ ndị ọrụ ịzụ ahịa ọsọ maka izi ezi. N'ihi na a na-enye ikike nha nha n'ihu ọha n'okpuru MIT, Whisper ghọrọ ọkpụkpụ ndabere maka ọtụtụ ndegharị pọdkastị, ngwaọrụ ntinye okwu na ngwa olu ihe fọrọ nke nta ka ọ bụrụ otu abalị.
Nghọta nka nka
Whisper na-ekewa ọdịyo n'ime nkeji nkeji iri atọ, tụgharịa nke ọ bụla ka ọ bụrụ log-Mel spectrogram (ọwa ugboro 80), wee nye ya na ntinye ihe ntụgharị. Onye ngbanwe ahụ na-ebu amụma akara ederede n'onwe ya, na-eduzi akara ngosi pụrụ iche na-akọwapụta ọrụ ahụ (nsụgharị vs. ntụgharị asụsụ), asụsụ, na ma ọ ga-ebunye akara oge. Nke a multitask token-conditioning bụ aghụghọ aghụghọ: otu nha nha na-arụ ọtụtụ ọrụ dabere na akara ngosi ozugbo ewepụtara na mmalite nke ngbanwe.
Mmetụta atụmatụ
Nweta na iru
Ọ na-eme ka nnweta ya dịkwuo mma site na ndegharị, ịkọ akụkọ, na ntụgharị olu.
Ọnụ ego na mmefu ego
Ndị otu mgbasa ozi nwere ike ibubata ọdịyo a na-egbu maramara ngwa ngwa site na iji obere mmefu ego.
Ọsọ na ọnụ ọgụgụ
Sistemụ na-eche ihu ndị ahịa nwere ike hazie mkparịta ụka n'ọtụtụ buru ibu.
Ọdịnihu nke ịmata okwu ntanye
Amụma kpalitere ụdị ngwa ngwa dị ka Whisper.cpp, ntasị ọsọ ọsọ, na ụdị dị iche iche na-agba ọsọ ozugbo na ekwentị na laptọọpụ. Na-atụ anya ụdị mgbasa ozi siri ike (obere oge) dị iche iche, ejikọta ọnụ okwu ka mma n'akụkụ ya, yana ịrụ ọrụ siri ike na asụsụ ndị dị ala. Ka ọdịyo ngwaọrụ AI na-etolite, ụdị ụdị Whisper dị fechaa ga-eme ka okwu dị ndụ, ndetu nzuko na ngwaọrụ nnweta kpamkpam na-anọghị n'ịntanetị, na-echekwa nzuzo ka ọ dabara na ọkwa igwe ojii.
Mmejuputa n'ezie n'ụwa
Na-emepụta transcript nke nwere ike ịchọ na akpaaka maka pọdkastị na vidiyo YouTube
Na-eme ka ngwa ndetu nzukọ dị ndụ na-ewepụta nchịkọta sitere na ụda Zoom ma ọ bụ Otu Otu
Ịsụgharị ajụjụ ọnụ n'asụsụ mba ofesi ozugbo na ederede Bekee maka ndị nta akụkọ
Iwulite ngwaọrụ nnweta na-achịkwa olu yana nkọwa okwu maka ndị ọrụ na-enweghị ike pịnye
Ihe ize ndụ & okporo ụzọ nche
Iji olu eme ihe na ihe egwu mpụta ga-abawanye mgbe nkwenye na-efu.
Izi ezi nwere ike ịdaba n'ofe ụda olu, olumba ma ọ bụ gburugburu mkpọtụ.
Enwere ike imehie ọdịyo sịntetik dị ka ezigbo okwu na-enweghị akara doro anya.
Map mmejuputa
Nweta nkwenye doro anya maka ijide olu, imechi, na ijigharị.
Nwale ogo n'ofe ndị na-ekwu okwu dị iche iche yana ọnọdụ ndabere.
Kọwaa mgbe mmadụ ga-enyocha ma ọ bụ kwado nsonye.
Deba aha ọdịyo sịntetik ma debe ndekọ ihe ndekọ maka ịza ajụjụ.
Nọgide na-eme nchọpụta
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
Ntuziaka na-esote
Nghọta mmetụta mmetụta okwu
Ajụjụ a na-ajụkarị
What is Whisper Speech Recognition?
Whisper bụ OpenAI sistemụ njirimara okwu akpaka nke mepere emepe nke na-atụgharị ọdịyo ka ọ bụrụ ederede n'ofe asụsụ 90+. Ọ dị mkpa n'ihi na ọ wetara onye ọ bụla ogo ederede mmadụ nso n'efu, na-arụ ọrụ siri ike na ụda olu, mkpọtụ ndabere, na nka nka.
Ihe dị ka elekere ole ka a zụrụ Whisper na ọdịyo?
A zụrụ Whisper n'ihe dị ka awa 680,000 nke ọtụtụ asụsụ, ọdịyo multitask anakọtara na webụ, nnukwu data dị iche iche.
Kedu ihe nnochite anya n'etiti Whisper na-agbakọ site na ọdịyo raw tupu koodu mkpuchi?
Whisper na-atụgharị nchikota ọdịyo nke nkeji iri atọ ọ bụla ka ọ bụrụ ọwa 80 log-Mel spectrogram, nke ihe ngbanwe nke Transformer na-eme.
Kedu ka otu ụdị Whisper si arụ ọtụtụ ọrụ dị ka ndegharị na ntụgharị asụsụ?
Ọnọdụ ntanye na akara pụrụ iche na mmalite nke ngbanwe nke na-agwa ya asụsụ, ọrụ, na ma ọ ga-ebunye akara oge.
Kedu ihe owuwu akwara ozi Whisper na-eji?
Whisper bụ ihe ngbanwe ihe ngbanwe: ihe ngbanwe na-agụ spectrogram na ihe ngbanwe na-ewepụta akara ederede na-akpaghị aka.
Kedu ihe kpatara eji ewere Whisper ka ọ siri ike ma e jiri ya tụnyere sistemụ ASR mbụ?
Ọzụzụ n'ọtụtụ dị iche iche, ọdịyo ụwa n'ezie (ụda olu, mkpọtụ, crosstalk) nyere Whisper ike siri ike na igbe na-enweghị nhazi nke ngalaba ọ bụla.