Vokoder WaveGAN paralel
Parallel WaveGAN vocodeur neuronal bu gaaw la buy soppi spectrogram mel ci forme onde audio bu ñor, di jëfandikoo GAN bu ndaw, di defar misaal yépp benn yoon.
Résumé
It matters because it gives near-real-time, high-quality speech with a compact model.
Plongeur bu xóot
Vocoder mooy wàll wi mujj ci pipeline TTS: dafay soppi kàrtu màndarga akustik (dafay faral di nekk mel-spectrogram) mu nekk onde son bi ngay dégg. WaveGAN parallel, Yamamoto, Song, ak Kim ñoo ko xalaat ci 2019, def lii ak benn generatër bu amul benn xeetu WaveNet buñ tàggat muy reso buy defar ay noon. Duñu wax benn misaalu audio benn yoon ni WaveNet bu njëkk bi, waaye dafay defar jëmmu onde bi yépp ci paralel, muy gëna gaaw. Recetam bu am solo bi dafa boole ñàkkaale ak ñàkkaale Fourier transformation (STFT) bu gàtt, suko defee model bi méngoo ak siñaal dëgg bi ci diir yu bari ak ci eskaalu fréquence. Lépp soo ko boolee mu am generatër bu ndaw (lu tollu ci 1.4 milioŋ ciy parametre) buy daw lu bari yoon lu gëna gaaw ci GPU.
Gis-gis xarala
Generatër bi ab reso bu yaatu la bu lalu ci mel-spectrogram ak ab dugal bruit, di xayma bruit bi ak ay man-man ci misaal yi. Taggat yaram dafay wàññi ñàkka am resolusioŋ STFT yu bari, ñu xayma ko ci méngale spectrogram magnitude ci dayo FFT yu bari ak guddaayu hop, ak ñàkkaale bu bawoo ci benn discriminateur buy àtte dëgg. Term STFT dafay dakkal ak gaawal tàggat yaram, jàpp detay yu rafet ak jëmmu spectral bu yaatu te du am distillation.
njeextalu pexe
Dugg ak yegg
Dafay gëna yombal jëfandikoo gi jaaraleko ci transkripsioŋ, nettali ak interfaasu baat.
Njëgg ak budget
Ekipu mejaa yi mën nañu yónnee audio bu leer ci anam wu gëna gaaw te seen xaalis gëna néew.
Gaawaay ak yaatuwaay
Sistem yiy jàkkarloo ak kiliyaan bi mën nañu def waxtaan ci anam wu gëna yaatu.
Ëlëgu vokoder WaveGAN paralel
Parallel WaveGAN jàppale ci taxawal vocoders GAN muy default praktis, ak résolution STFT yu bari yi ñu ñàkk leegi ñu ngi feeñ ci ñi ko wuutu lu melni HiFi-GAN ak sistem streaming yu bari. Trajectoire bi dafay wane vocoder yu gëna ndaw, yu gëna néew latency ngir assistant yi ci aparey yi, aparey yiy déglu, ak soppi baat ci saasi, boole ci vocoder universel yuy yamale kàddukat yuñ gisul. Xaarandi lëkkaloo bu gëna dëgër ak TTS end-to-end ak jëfandikoo bu baax ci puce mobile ak yuñ samp.
Doxal ci àdduna dëgg
Wax jiy génne ci jamono dëgg ci assistant baat mobile, fu latency ak dayo model bi am solo
Dafay nekk generatëru forme onde boole ci model akustik yu melni Tacotron 2 wala FastSpeech
Bind-ci-kaddu ci aparey bi ngir jumtukaayi yombal jëfandikoo gi mënul yéem ci niir yi
Sistem yuy soppi baat yiy soppi spectrogram yi ñu soppi leen ñu nekk audio buy sone
Risk yi ak balustrade yi
Jëfandikoo baat ci anam wu jaarul yoon ak niru ak nit dafay gëna yokk sudee nanguwul.
Jaar-jaar mën na wàññeeku ci aksan yi, dialect yi wala barab yu bari xumbaay.
Audio synthetik mën nañu ko jaawale ak wax ju dëggu sudee amul etiket bu leer.
Roadmap ngir samp gi
Wutal ndigal bu leer ngir jàpp baat bi, klone ko ak jëfandikoowaat ko.
Saytu kalite ci kàddukat yu bari ak anam yu bari ci ginaaw.
Mandargal kañ la nit wara xoolaat wala nangu ay génne.
Etiketu audio synthetik te nga denc dokimaa ci fimu bawoo ngir mëna lim.
Weyal di banneexu
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 Parallel WaveGAN 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
Gis bi ci topp
MelGAN vocodeur buy defar
Laaj yi ñuy faral di laaj
What is Parallel WaveGAN Vocoder?
Parallel WaveGAN vocodeur neuronal bu gaaw la buy soppi spectrogram mel ci forme onde audio bu ñor, di jëfandikoo GAN bu ndaw, di defar misaal yépp benn yoon. Dafa am solo ndax dafay joxe kàddu yu am kalite ci jamono dëgg, ak model bu dëgër.
Luy liggéeyu vocoder bu melni WaveGAN paralel?
Vocoder mooy etape TTS bi mujj biy boole ondes son yi ci man-mani akustik yu melni mel-spectrogram.
naka la WaveGAN parallel di defaree ay misaali audio buñu ko méngale ak WaveNet bu njëkk bi?
Parallel WaveGAN du autoregressive, dafay defar forme onde bi yépp ci benn yoon, du muy génne misaal par misaal, moo tax mu gëna gaaw.
Ban perte moo gëna am solo ci Parallel WaveGAN ginaaw perte biy xeex?
Dafay boole ñàkkaale bu ànd ak ñàkkaale STFT bu bari dayo buy méngale magnitude spectrogram ci eskaal yu bari.
Ban tabax la generatëru WaveGAN paralel bi di jëfandikoo?
Generatër bi reso bu niruwaale WaveNet la, ñu tabax ko ci ay konvolusioŋ yu yaatu, lalu ci spectrogram mel ak bruit.
Lu tax WaveGAN paralel neexa jëfandikoo?
Ak lu tollu ci 1.4 milioŋ ciy paramet, dafa tuuti te dafay daw lu bari yoon ci waxtu dëgg, baaxna lool ci jëfandikoo ci aparey bi.