Okwu ngwa ngwa na TTS na-abụghị nke autoregressive
FastSpeech na-ewepụta spectrogram okwu dum n'otu n'otu karịa otu etiti n'otu oge, na-eme ka njikọ ahụ dị ngwa ngwa ma kwụsie ike.
Nchịkọta
It solved the slow, error-prone generation that plagued earlier autoregressive models like Tacotron.
Ime miri emi
Ụdị TTS ndị mbụ dị ka Tacotron 2 bụ ndị na-emegharị anya: ha na-ebu amụma ụda olu ọ bụla nwere ọnọdụ na nke gara aga, nke na-adị ngwa ngwa ma na-adịkarị mfe mafe ma ọ bụ okwu ugboro ugboro mgbe nlebara anya na-ezighị ezi. FastSpeech, nke __AIU_PROTECTED_5_ na Mahadum Zhejiang webatara na 2019, tụgharịrị nke a site na ịkọ okpokolo agba niile n'otu oge. Netwọk mgbasa ozi nke dabere na ntụgharị na-ewe fọnịm, na-ebu amụma n'ụzọ doro anya ogologo oge ekwentị ọ bụla ga-adịru site na iji ogologo nhazi ogologo, wee gbasaa usoro ahụ gaa na ọnụọgụ okpomoku ziri ezi tupu ịmepụta spectrogram n'otu ngafe. FastSpeech 2 kwalitere na nke a site n'ịkọ ọkwa ọkwa na ike yana site na ọzụzụ oge ebumnuche site na itinye mmanye kama ịwepụ ha site na ụdị onye nkuzi dị nwayọ, na-enyekwu okwu okike na njikwa.
Nghọta nka nka
Isi aghụghọ bụ ogologo regulator. N'ihi na ederede na ọdịyo nwere ogologo dị iche iche, FastSpeech na-ebu amụma ogologo oge maka fọnịm ọ bụla ma na-ekwughachi ọnọdụ ezoro ezo nke fọnịm ahụ ọtụtụ oge iji dakọọ ogologo spectrogram. Nhazi a doro anya na-anọchi anya adịghị ike. Ịmepụta etiti ọ bụla n'otu n'otu pụtara oge ntinye aka na-adabere na ogologo nkebiokwu, na iwepụ akaghị aka na-ekpochapụ mperi nke ịwụ na ikwughachi okwu.
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 FastSpeech na TTS na-abụghị autoregressive
Njikọ na-abụghị autoregressive ugbu a bụ ndabara maka mmepụta TTS n'ihi na ọ dị ngwa ngwa, siri ike, na njikwa. Sistemu ọdị n'ihu na-aga n'ihu na njikwa prosody kacha mma, mgbasa ozi dị ala maka ngwa dị ndụ, yana ụdị ngwụcha ruo na njedebe nke na-amapụ spectrogram etiti kpamkpam. Ụdị mgbasa ozi na-adabere na ụdị na-abụghị autoregressive na-arị elu, na-agwakọta FastSpeech's parallelism with siri ike generative quality, ebe doro anya na njikwa oge na-eji kpọrọ ihe maka edizi, na-egosipụta olu ngwaahịa.
Mmejuputa n'ezie n'ụwa
Ngwa igodo ozugbo na-emepụta mkpali olu na-atụgharị n'otu ntabi anya site na iji nhazi ụdị FastSpeech.
Sistemụ ọrụ ndị ahịa-IVR na-atụgharị ederede dị ike ka ọ bụrụ okwu n'ọtụtụ na-enweghị njehie mwụli okwu.
Ndị na-agụ ihuenyo nnweta na-ewepụta ngwa ngwa, okwu ntụkwasị obi maka akwụkwọ ogologo na ngwaike dị obere.
Ngwa ọdịnaya olu na-eme ka ndị okike tweak pitch na ọnụego ikwu okwu ozugbo, ekele maka ngwa ngwa ngwa ngwa na amụma ike nke FastSpeech 2.
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 FastSpeech and Non-Autoregressive TTS 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
Mbe TTS Autoregressive Synthesis
Ajụjụ a na-ajụkarị
What is FastSpeech and Non-Autoregressive TTS?
FastSpeech na-ewepụta spectrogram okwu dum n'otu n'otu karịa otu etiti n'otu oge, na-eme ka njikọ ahụ dị ngwa ngwa ma kwụsie ike. Ọ doziri ọgbọ na-adịghị ngwa ngwa, nke na-enweghi ike imejọ nke mebiri ụdị autoregressive mbụ dị ka Tacotron.
Kedu ihe 'na-abụghị autoregressive' pụtara na ọnọdụ nke FastSpeech?
A na-emepụta fremu na-abụghị nke autoregressive n'otu n'otu n'otu ngafe, ọ bụghị n'otu n'otu na okpokolo agba gara aga.
Kedu nsogbu nke ụdị autoregressive dị ka Tacotron 2 ka FastSpeech na-ekwu kpọmkwem?
Nlebara anya na-eweghachite onwe ya nwere ike ịmekọrịta nke ọma, na-ebute okwu awụsara ma ọ bụ ugboro ugboro, na ngbanwe usoro na-adị ngwa ngwa; FastSpeech na-edozi ma.
Gịnị bụ ọrụ nke 'ogologo regulator' na FastSpeech?
Onye na-ahazi ogologo oge na-agbasa njirimara phoneme site na ogologo oge ha buru amụma, na-ahazi usoro ederede dị mkpụmkpụ na spectrogram dị ogologo.
Kedu ihe FastSpeech 2 gbakwunyere ma e jiri ya tụnyere nke mbụ FastSpeech?
FastSpeech 2 na-ebu amụma pitch na ume ma na-eji oge ntinye mmanye kama ịbụ onye nkuzi dị nwayọọ, na-eme ka ọdịdị na njikwa dịkwuo mma.
Kedu ihe kpatara oge ntinye aka nke FastSpeech ji eto ogologo oge?
N'ihi na ọgbọ dị n'otu n'otu, ịgbakwụnye okpokolo agba anaghị amụba usoro n'usoro dịka ngbanwe autoregressive si eme.