UnivNet Multi-Mkpebi Vocoder
UnivNet bụ GAN vocoder nke na-ekpe ikpe na-emepụta ọdịyo site na iji ọtụtụ spectrogram gbakọrọ na mkpebi STFT dị iche iche, na-amụba nkọwa zuru oke.
Nchịkọta
It aims to be a universal vocoder that generalizes well to unseen speakers and recording conditions.
Ime miri emi
UnivNet, nke Jang et al tụpụtara ya. na 2021, na-egbochi adịghị ike nke GAN vocoders na-ahụkarị: ugboro ugboro dị elu nke gbachiri emechi ma ọ bụ ihe arụrụ arụ. Ọnọdụ generator ya na mel-spectrograms zuru oke ma na-eji ọnọdụ mgbanwe mgbanwe ọnọdụ (LVC), ebe a na-ebu amụma kernel convolution na ofufe site na njirimara ntinye ka nzacha ahụ dabara na ọdịnaya mpaghara. Echiche isiokwu bụ multi-resolution spectrogram discriminator (MRSD): kama ikpe naanị ụdị waveform, UnivNet na-agbakọ ọtụtụ STFT nwere windo dị iche iche na nha hop ma na-agba ndị ịkpa ókè n'ogo spectrogram ahụ. Nke a na-akpali onye na-emepụta ọkụ ka ọ nweta ma ọmarịcha nkọwapụta na nhazi oge sara mbara nke ọma. N'ịbụ onye a zụrụ n'ọtụtụ ndị na-ekwu okwu, UnivNet na-ewepụta okwu okike maka olu ọ na-ahụtụbeghị n'oge ọzụzụ, na-enweta akara ya zuru ụwa ọnụ.
Nghọta nka nka
Ọnọdụ UnivNet na-agbanwe agbanwe na-ewepụta oke kernel ya ike site na njirimara mel na-edozi site na obere netwọkụ kernel-predictor, yabụ usoro oge ọ bụla na-eji nzacha na-eme ka ọ dị mma karịa kernel ekekọrịtara. Ejikọtara ya na ịkpa ókè spectrogram nke ọtụtụ mkpebi, nke na-agbasa ọtụtụ oge azụmaahịa n'otu oge, nke a na-elekwasị anya n'otu egwu dị elu ebe ndị vocoders GAN dị mfe na-agbaji ma ọ bụ na-akparị.
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 UnivNet Multi-Resolution Vocoder
UnivNet's multi-resolution spectrogram discrigram aghọọla ihe eji eme ihe na ngwugwu TTS ọgbara ọhụrụ yana emetụta sistemu dị ka BigVGAN na codec ọdịyo akwara. Na-atụ anya ka ihe zuru ụwa ọnụ, ọkà okwu-agnostic Framing ga-anọgide na-agbasawanye n'ebe ịbụ abụ olu, multilingualism njikọ, na bandwit zuru ezu 48 kHz audio, mgbe echiche kernel-emegharị na-agwa ndị na-arụ ọrụ nke ọma ụdị na ngwaọrụ na-aghaghị ijikwa dị iche iche olu na-enweghị onye ọ bụla na-ekwu okwu ọma nlegharị anya.
Mmejuputa n'ezie n'ụwa
Ọrụ TTS ọtụtụ ndị na-ekwu okwu nke ga-adarịrị ụda eke na olu adịghị adị na data ọzụzụ
Pipeline cloning olu ebe otu vocoder zuru ụwa ọnụ na-ejere ọtụtụ ndị na-ekwu okwu ebumnuche
Akwụkwọ ọdịyo nwere ntụkwasị obi dị elu na akụkọ pọdkastị chọrọ sibilance dị nkọ na ugboro dị elu
Vocoder azụ maka sistemu TTS ngwụcha ruo ọgwụgwụ nke na-ejikọ amụma spectrogram na igwe na-ebupụta ebili mmiri siri ike.
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 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
Ntuziaka na-esote
Vocoders akwara
Ajụjụ a na-ajụkarị
What is UnivNet Multi-Resolution Vocoder?
UnivNet bụ GAN vocoder nke na-ekpe ikpe na-emepụta ọdịyo site na iji ọtụtụ spectrogram gbakọrọ na mkpebi STFT dị iche iche, na-amụba nkọwa zuru oke. Ọ na-achọ ịbụ vocoder zuru ụwa ọnụ nke na-agbasa nke ọma na ndị na-ekwu okwu na-adịghị ahụ anya na ọnọdụ ndekọ.
Kedu ihe bụ njirimara mbinye aka nke ịkpa oke UnivNet?
UnivNet's multi-resolution spectrogram discriminator na-enyocha ọtụtụ STFT nwere windo dị iche iche na nha hop iji weghara azụmaahịa oge dị iche iche.
Kedu nsogbu dị na mbụ GAN vocoders UnivNet na-ezubere iche?
Site n'ịkpa oke n'ofe ọtụtụ mkpebi spectrogram, UnivNet na-eme ka nkọwapụta ọnụọgụ dị elu nke ndị ụda olu GAN dị mfe na-ejikarị eme ihe.
Kedu ihe bụ mgbanwe mgbanwe ọnọdụ (LVC) na UnivNet?
LVC na-eji netwọk kernel-predictor ka ọnọdụ ọ bụla na-etinye ihe nzacha na-emegharị ọdịnaya sitere na mel-spectrogram na-edozi ahụ.
Kedu ihe kpatara eji kọwaa UnivNet dị ka ụda olu zuru ụwa ọnụ?
N'ịbụ onye a zụrụ n'ọtụtụ ndị na-ekwu okwu, UnivNet na-emepụta okwu okike ọbụna maka olu ọ na-ahụtụbeghị na ọzụzụ, ya mere zuru ụwa ọnụ.
Kedu ka ịkpa ókè spectrogram nke multi-resolution si enyere onye na-emepụta ihe aka?
Mkpebi STFT dị iche iche na-emesi ụdị azụmaahịa oge dị iche iche, yabụ imekọ ha niile na-akpali onye na-ere ọkụ na nkọwa zuru oke na nhazi ya.