MelGAN Vocoder
MelGAN nijwi ryuzuye rya GAN rishingiye kumajwi rihindura mel-spectrogramu muburyo bwamajwi mbisi mumajwi imwe yihuta.
Incamake
It mattered because it proved high-quality, non-autoregressive speech synthesis could run hundreds of times faster than real time on a GPU.
Kwibira cyane
MelGAN, yatangijwe na Kumar n'abandi. muri 2019, itanga amajwi nta gahoro gahoro-by-sample ikoreshwa na WaveNet. Imashini itanga amashanyarazi ni ihuriro ryimyanya ndangagitsina yerekana urugero rwa mel-spectrogramu (mubisanzwe imirongo 80 yumurongo) kugeza ku gipimo cyerekana amajwi, hamwe nibisigara bisigaye bifashisha imyanzuro yagutse kugirango bagure umurima wakira. Agashya k'ingenzi kwari uguhugura hamwe n'ivangura ryinshi rikorera ku munzani utandukanye w'amajwi (umwimerere wambere wongeyeho verisiyo yamanutse), buri wese akareba Windows yuzuye. Igihombo gihuye nikigereranyo kigereranya ibikorwa bivangura hagati yijwi ryukuri nimpimbano, bigahindura imyitozo ya GAN. Icyitegererezo ni gito kubijyanye na neural-majwi kandi ikora byihuse kuruta igihe nyacyo ndetse no kuri CPU, bigatuma iba ingirakamaro kubishyizwemo no ku gikoresho inyandiko-ku-mvugo.
Ubushishozi
MelGAN ivangura ryinshi ikoresha imiyoboro itatu isa ireba amajwi yuzuye, igice, na kimwe cya kane, buri cyiciro cyo gufata kumurongo utandukanye. Icy'ingenzi, MelGAN yishingikiriza ku gihombo gihuye (intera ya L1 hagati yikarita yerekana ivangura ryerekana amajwi nyayo yakozwe n’amajwi) aho kuba igihombo cyerekana kwiyubaka, gishishikariza generator guhuza imibare nyayo y’amajwi.
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 ha MelGAN Vocoder
MelGAN yabibye umuryango wabatoye GAN. Abasimbuye, HiFi-GAN na UnivNet, bakomeje uburyo bwihuse butari ubwikorezi ariko bongeraho ivangura ryibihe byinshi kandi bikemura byinshi kugirango bisukure cyane. Ubwubatsi butuye ku bikoresho no gutambutsa TTS aho ubukererwe nubunini bwikitegererezo bifite akamaro, kandi ibitekerezo byayo bivangura bikomeje kugira ingaruka kuri codec ya neural na sisitemu yo kubyara imiziki aho imyitozo yo guhangana iteza imbere ireme ryimyumvire.
Gushyira mu bikorwa Isi
Ku gikoresho cyanditse-ku-mvugo mu bafasha ba mobile aho ijwi rito, ryihuta ryirinda ingendo zuzenguruka
Ijwi-nyaryo ryo guhindura imiyoboro ihindura imvugo ya mel-spectrogramu mumajwi yagenewe
Imikino na animasiyo ibikoresho bihuza ibiganiro byimiterere biva kuri spekrogramu yakozwe hamwe nubukererwe buke
Ibyingenzi byubushakashatsi kumajwi GAN, aho MelGAN yibiranga-bihomba byongeye gukoreshwa mumuziki no kubyara amajwi
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 MelGAN Generative 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
Vocoder ya DiffWave
Ibibazo bikunze kubazwa
What is MelGAN Generative Vocoder?
MelGAN nijwi ryuzuye rya GAN rishingiye kumajwi rihindura mel-spectrogramu muburyo bwamajwi mbisi mumajwi imwe yihuta. Byari bifite akamaro kuko byagaragaye ko bifite ireme ryiza, bitavugwaho rumwe imvugo ishobora gukora inshuro magana byihuse kuruta igihe nyacyo kuri GPU.
Ni ubuhe butumwa MelGAN ihindura mu majwi mbisi?
MelGAN ni amajwi: ifata imiterere ya acoustic igereranya, mel-spectrogram, kandi ikomatanya imiterere ihuye nayo.
Kuki MelGAN yihuta cyane kuruta WaveNet?
WaveNet itanga ingero imwe icyarimwe; Imashini itanga imbaraga za MelGAN itanga umurongo wose muburyo bumwe.
Ni ubuhe buryo bwihariye bwo kuvangura MelGAN yazanye?
MelGAN ikoresha ivangura ryinshi risuzuma imiterere yumurongo wumwimerere hamwe na verisiyo yamanutse kugirango ifate imiterere kumunzani itandukanye.
Niki gihombo gifasha guhagarika imyitozo yo guhangana na MelGAN?
Igihombo-gihuye nigihombo kigabanya intera ya L1 hagati yikarita iranga ivangura kumajwi nyayo kandi yakozwe, kuyobora generator no guhugura amahugurwa.
Nigute generator ya MelGAN yongera umurima wakira?
Ibisigisigi bisigaye bifite imyanzuro yagutse byagura umurima wakira neza, ukareka moderi ya generator ikagereranya imiterere yigihe gito.