Ubuyobozi bwa Audio AI

Gutandukanya disikuru

Gutandukanya disikuru asubiza ikibazo "ninde wavuze ryari?" mugabanye amajwi yafashwe mubice byanditseho indangamuntu.

2 min somaIbiherutse kuvugururwa

Incamake

It turns a single stream of mixed voices into a timeline showing exactly which person was talking at each moment.

Kwibira cyane

Diarisation itunganya amajwi mubyiciro. Ubwa mbere, ibikorwa byerekana amajwi bisanga uturere tuvuga. Imvugo noneho yaciwemo ibice bigufi, kandi buri gice gihindurwamo icyerekezo kirekire-cyiswe icyuma gishyiramo (amateka ya i-vectors cyangwa x-vectors, ubu ubusanzwe ni imitekerereze ya ECAPA-TDNN). Intambwe yo guhuriza hamwe (agglomerative clustering cyangwa spectral clustering) amatsinda yibice hamwe nibisobanuro bisa mubavuga, akenshi utazi umubare wabavuga mbere. Hanyuma, imipaka iratunganijwe kandi imvugo irenze ikemurwa. Icy'ingenzi, diarisation ntabwo ikeneye kumenya abantu abo ari bo; igenera gusa ibirango bitazwi nka "Umuvugizi 1" na "Umuvugizi 2." Ukuri gupimwa hamwe na Diarisation Error Rate (DER), ihuza imvugo yabuze, gutabaza ibinyoma, no kwitiranya abavuga.

Ubushishozi

Amayeri yibanze ni disikuru ishiramo: urusobe rw'imitsi yatojwe kuburyo amashusho avuye kumuntu umwe agwa hamwe mumwanya wa vector hamwe na clips zabantu batandukanye zigera kure. Clustering noneho ikora kuriyi mikorere aho kuba amajwi mbisi. Igezweho rya "end-to-end near diarisation" (EEND) risimbuza cluster hamwe numuyoboro umwe ukoresheje amahugurwa ya permisation-invariant, ikora imvugo ihuzagurika cyane kuruta imiyoboro ihuza gusa umuvugizi umwe umwe.

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.

Ejo hazaza h'umuvugizi Diarisation

Diarisation ni uguhinduranya hamwe na moderi ihuriweho hamwe isohora amagambo hamwe na labels ibirango mumurongo umwe, bigabanya kwibeshya. Tegereza uburyo bwiza bwo kuvuga neza, inama nini hamwe nabitabiriye amahugurwa benshi, hamwe nigihe nyacyo cyo gutondeka neza. Kwiyobora kwifata ryerekana amajwi hamwe nibimenyetso byinshi (kugenda iminwa, icyerekezo cyo kuva kuri mikoro ya mikoro) bizarushaho gukosorwa neza, mugihe diarisation yibikoresho bizamura ubuzima bwite mugukomeza amakuru yijwi.

Gushyira mu bikorwa Isi

Gukora disikuru yanditseho inyandiko zinama zubucuruzi mubikoresho nka Otter.ai cyangwa Microsoft Amakipe

Gukora "ninde wavuze iki" igihe cya podcast na software ikosora

Kwerekana guhamagarwa-hagati byafashwe amajwi kugirango utandukanye abakozi nabakiriya bahindure isesengura ryiza

Kubaka icyumba cyurukiko no kubitsa amajwi kugirango buri muvugizi atangwe neza

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

1

Shaka uruhushya rusobanutse rwo gufata amajwi, gukoroniza, no gukoresha.

2

Ikizamini cyiza mubiganiro bitandukanye hamwe nuburyo bwimbere.

3

Sobanura igihe umuntu agomba gusuzuma cyangwa kwemeza ibisubizo.

4

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 Speaker Diarization quiz

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

Tangira ikibazo

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

Ubuyobozi bukurikira

ECAPA-TDNN Kumenyekanisha Umuvugizi

Ibibazo bikunze kubazwa

What is Speaker Diarization?

Gutandukanya disikuru asubiza ikibazo "ninde wavuze ryari?" mugabanye amajwi yafashwe mubice byanditseho indangamuntu. Ihindura umurongo umwe wamajwi avanze mugihe cyerekana neza umuntu yavugaga buri mwanya.

Ni ikihe kibazo cy'ibanze disikuru igamije gusubiza?

Diarisation igabanya amajwi numuvugizi mugihe, asubiza "uwavuze iyo" aho kwandukura amagambo ubwabo.

Umuvugizi ushiramo iki?

Umuvugizi winjizamo ni uburebure buringaniye aho clips ziva kumuntu umwe zegeranye hamwe, bigafasha guhuza kubiranga.

Nibihe bipimo bisanzwe bikoreshwa mugusuzuma sisitemu yo gutandukana?

DER ikomatanya imvugo yabuze, impuruza zitari zo, hamwe no kwitiranya abavuga mu ijanisha rimwe, bigatuma ibipimo bisanzwe bya diarisation.

Ese diarisation isanzwe ikeneye kumenya amazina nyayo y'abavuga hakiri kare?

Diarisation cluster amajwi kandi igenera ibirango bitazwi; ntibisaba kumenya abo mubyukuri abo ari bo mwizina.

Ni ukubera iki moderi ya diarisation iherezo-iherezo (EEND) ihabwa agaciro hejuru y'imiyoboro ihuriweho gusa?

Moderi ya EEND ikoresha imyitozo ya permisation-idahinduka kugirango yerekane mu buryo butaziguye abavuga rikijyana, ikora imvugo irenga igabanya umuvugizi-icyarimwe.