UnivNet Multi-Resolution Vocoder
UnivNet numu majwi wa GAN abacamanza bakoze amajwi bakoresheje spekrogramu nyinshi zabazwe mubyemezo bitandukanye bya STFT, bikarishye birambuye.
Incamake
It aims to be a universal vocoder that generalizes well to unseen speakers and recording conditions.
Kwibira cyane
UnivNet, yatanzwe na Jang n'abandi. muri 2021, ikemura intege nke zisanzwe zijwi rya GAN: zivanze cyangwa ibihangano byuzuye imitwaro myinshi. Imashini itanga amashanyarazi kuri bande yuzuye ya mel-spectrograms kandi ikoresha ahantu-ihindagurika ihindagurika (LVC), aho intangiriro zo guhinduranya ziteganijwe kuguruka ziva mubintu byinjijwe bityo akayunguruzo gahuza nibirimo byaho. Igitekerezo cyumutwe nigisubizo cyinshi cyerekana ivangura (MRSD): aho guca imanza zifatika gusa, UnivNet ibara STFT nyinshi zifite idirishya ritandukanye hamwe nubunini bwa hop kandi ikoresha ivangura kuri ubwo bunini bwa spekrogramu. Ibi bisunika generator kugirango ibone ibisobanuro byiza byose hamwe nuburyo bwagutse burigihe. Yahuguwe kubavuga benshi, UnivNet itanga imvugo karemano kumajwi itigeze ibona mugihe cyamahugurwa, ikabona ikirango cyayo rusange.
Ubushishozi
Ikibanza cya UnivNet gihinduka-gihindura imyumbati itanga uburemere bwimbaraga zivuye mumiterere ya mel ikoresheje urusobe ruto-ruteganya, bityo buri gihe intambwe ikoresha neza ibiyikubiyemo-bihuza na filteri aho kuba intangiriro ihuriweho. Ufatanije nivangura ryinshi rya spekrogramu ivangura, ikurikirana inshuro nyinshi inshuro nyinshi gucuruza icyarimwe, ibi byibanda kumurongo wihuta cyane aho amajwi yoroshye ya GAN akunda kuvuga nabi cyangwa hum.
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.
Kazoza ka UnivNet Multi-Resolution Vocoder
Ivangura rya UnivNet ryinshi-ivangura ryabaye ikintu gisanzwe mubice bya TTS bigezweho kandi bigira ingaruka kuri sisitemu nka BigVGAN na codecs zamajwi. Witege kuri bose, abavuga-agnostic gushiraho kugirango bakomeze kwaguka bagana amajwi yo kuririmba, guhuza indimi nyinshi, hamwe numuyoboro mugari wa 48 kHz amajwi, mugihe igitekerezo cyo guhuza n'imikorere-kernel kimenyesha imikorere yibikoresho bigomba gukoreshwa amajwi atandukanye nta muvugizi uhuza neza.
Gushyira mu bikorwa Isi
Serivisi nyinshi za TTS zigomba kumvikana kumajwi atagaragara mumahugurwa
Imiyoboro ya cloni yijwi aho ijwi rimwe ryogukorera abantu benshi bavuga
Igitabo cyizerwa cyane cyamajwi hamwe na podcast inkuru ikenera crisp sibilance na frequency nyinshi
Inyuma ya majwi ya sisitemu ya TTS iherezo-iherezo ihuza sisitemu yo guhanura hamwe na generator ikomeye.
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 UnivNet Multi-Resolution Vocoder 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
Vocoders
Ibibazo bikunze kubazwa
What is UnivNet Multi-Resolution Vocoder?
UnivNet numu majwi wa GAN abacamanza bakoze amajwi bakoresheje spekrogramu nyinshi zabazwe mubyemezo bitandukanye bya STFT, bikarishye birambuye. Ifite intego yo kuba amajwi rusange asanzwe neza kubavuga batagaragara hamwe nuburyo bwo gufata amajwi.
Ni ibihe bintu biranga umukono wa UnivNet ivangura?
UnivNet ikemura byinshi-ivangura ivangura isesengura STFT nyinshi zifite idirishya ritandukanye hamwe nubunini bwa hop kugirango ifate ibihe bitandukanye-byacurujwe.
Ni ikihe kibazo mu majwi ya mbere ya GAN UnivNet yibandaho?
Mu kuvangura imyanzuro myinshi ya spekrogramu, UnivNet itezimbere umurongo-mwinshi urambuye amajwi yoroshye ya GAN abavuga kenshi.
Ni ubuhe buryo buhinduka (LVC) muri UnivNet?
LVC ikoresha urusobekerane rwimikorere kuburyo buri kibanza gikoresha ibintu-bihuza na filteri ikomoka kuri conditioning mel-spectrogram.
Kuki UnivNet isobanurwa nkijwi rusange?
Yahuguwe kubavuga benshi, UnivNet ihuza imvugo karemano ndetse nijwi ritigeze rihura mumahugurwa, bityo kwisi yose.
Nigute ivangura ryinshi rya spekrogramu ivangura rifasha generator?
Imyanzuro itandukanye ya STFT ishimangira ibihe bitandukanye-inshuro zicuruzwa, bityo guhuza byose bigasunika generator muburyo burambuye nimiterere.