Ọkpụkpọ okwu okwu
Ọdịda ụda okwu na-aza ajụjụ a "onye kwuru mgbe?" site n'ikewa ndekọ ọdịyo gaa n'akụkụ nke ejiri njirimara ọkà okwu kpọọ.
Nchịkọta
It turns a single stream of mixed voices into a timeline showing exactly which person was talking at each moment.
Ime miri emi
Ọdịda ụda na-ahazi ọdịyo na nkebi. Nke mbụ, nchọpụta ọrụ olu na-achọpụta mpaghara okwu. A na-egbutu okwu ahụ ka ọ bụrụ nkebi dị mkpirikpi, a na-atụgharị akụkụ nke ọ bụla ka ọ bụrụ vector nwere ogologo ogologo a na-akpọ okwu embedding (akụkọ ihe mere eme i-vectors ma ọ bụ x-vectors, ugbu a na-abụkarị ihe ntinye akwara dị ka ECAPA-TDNN). Nzọụkwụ nchịkọta (agglomerative clustering or spectral clustering) ngalaba nwere ihe ntinye yiri ya n'ime ọkà okwu, mgbe mgbe na-amaghị ọnụọgụ ndị ọkà okwu n'ọdịnihu. N'ikpeazụ, a na-anụcha ókèala ma dozie okwu ndị jikọtara ọnụ. N'ụzọ dị mkpa, diaarization adịghị mkpa ịmara ndị mmadụ bụ aha; ọ na-ekenye aha ndị na-amaghị aha dịka "Ọkà Okwu 1" na "Ọkà Okwu 2." A na-eji ọnụ ọgụgụ Error Error (DER) tụọ nkenke nke ọma, nke jikọtara okwu efu, mkpu ụgha, na mgbagwoju anya ọkà okwu.
Nghọta nka nka
Isi aghụghọ bụ ntinye okwu okwu: netwọkụ akwara a zụrụ azụ ka vidiyo si n'otu onye ahụ wee dakọ ọnụ na oghere vector na vidiyo sitere na ndị dị iche iche na-agbada n'ebe dị anya. Ụyọkọ na-arụkwa ọrụ na ntinye ndị a karịa ọdịyo nkịtị. Oge a "njedebe na njedebe akwara ozi" (EEND) na-eji otu netwọk dochie nchịkọta site na iji ọzụzụ na-agbanwe agbanwe, nke na-ejikwa okwu na-agafe agafe nke ọma karịa nchịkọta-naanị pipeline nke na-ewere otu ọkà okwu n'otu oge.
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 Ọdịnihu Ọkà Okwu
Ntugharị ntugharị na-ejikọta na ndenye aha n'ụdị ejikọtara ọnụ nke jikọtara ọnụ na-ewepụta okwu na akara okwu n'otu ngafe, na-ebelata mkpokọta njehie. Na-atụ anya njikwa okwu dị n'otu n'otu ka mma, nnukwu nzukọ yana ọtụtụ ndị sonyere na mgbasa ozi ozugbo maka nkọwa okwu dị ndụ. Nnọchite anya ọdịyo nke onye na-ahụ maka onwe ya na akara multimodal (mmegharị egbugbere ọnụ, ntụzịaka nke mbata site na nhazi igwe okwu) ga-amụba izi ezi, ebe ịdekọ ngwaọrụ ga-emeziwanye nzuzo site na idobe data olu mpaghara.
Mmejuputa n'ezie n'ụwa
Ịmepụta transcript nke nzukọ azụmahịa akpọrọ ọkà okwu n'ime ngwaọrụ dịka Otter.ai ma ọ bụ Microsoft Otu
Na-ewepụta "onye kwuru gịnị" usoro iheomume maka pọdkastị na ngwanrọ ndezi ajụjụ ọnụ
Na-edekọ ndekọ etiti oku iji kewapụ ndị nnọchite anya na ntụgharị ndị ahịa maka nyocha dị mma
Ịhazi ọnụ ụlọ ikpe na ntinye ụda ka e wee kwupụta nkwupụta okwu ọ bụla nke ọma
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 Speaker Diarization 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
Nnabata Ọkà Okwu ECAPA-TDNN
Ajụjụ a na-ajụkarị
What is Speaker Diarization?
Ọdịda ụda okwu na-aza ajụjụ a "onye kwuru mgbe?" site n'ikewa ndekọ ọdịyo gaa n'akụkụ nke ejiri njirimara ọkà okwu kpọọ. Ọ na-atụgharị otu iyi nke olu agwakọtara ka ọ bụrụ usoro iheomume na-egosi kpọmkwem onye na-ekwu okwu n'oge ọ bụla.
Speaker diarization is often summarized by which question?
Nkewa okwu na-ekesa ọdịyo site n'ọkà okwu ka oge na-aga, na-aza "onye kwuru mgbe" kama ịdegharị okwu ndị ahụ n'onwe ha.
Gịnị bụ agbakwunyere ọkà okwu?
Ntinye okwu okwu bụ vector nwere ogologo ogologo ebe obere vidiyo sitere na otu onye nọ ọnụ, na-eme ka nchịkọta site na njirimara.
Kedu metrik a na-ejikarị iji nyochaa sistemu diariation?
DER na-ejikọta okwu efu, mkpu ụgha, na mgbagwoju anya ọkà okwu n'ime otu pasent, na-eme ka ọ bụrụ metric diarization ọkọlọtọ.
Ọ dị mkpa ka edepụta akwụkwọ akụkọ ọkọlọtọ mara ezigbo aha ndị ọkà okwu n'ọdịnihu?
Ụyọkọ edetu ụyọkọ na-achịkọta olu ma na-ekenye aha ndị na-amaghị aha; ọ dịghị achọ ịmata ndị bụ́ aha ndị mmadụ n'ezie.
Kedu ihe kpatara eji eji ụdị ọnụọgụ ọnụọgụgụ akwara (EEND) na ngwụcha ruo na ngwụcha kpọrọ ihe karịa ọkpọkọ-naanị ọkpọkọ?
Ụdị EEND na-eji ọzụzụ na-adịghị agbanwe agbanwe na-eme ka ọtụtụ ndị na-ekwu okwu na-eṅomi kpọmkwem, na-ejikwa okwu na-emekọ ihe ọnụ na-emebi nchịkọta otu onye na-ekwu okwu n'otu oge.