EnCodec Audio mkpakọ
EnCodec bụ Meta's codec ụda olu nwere ntụkwasị obi dị elu nke na-akpakọ okwu na egwu n'ụdị bitrate dị ala nke nwere ụdị njiri mara mma.
Nchịkọta
It matters because it underpins modern generative audio systems and ships in open-source form for anyone to use.
Ime miri emi
Nke Meta AI weputara na 2022, EnCodec na-agbaso ụkpụrụ SoundStream nke ihe ngbanwe, vector quantizer residual vector (RVQ), na ihe ngbanwe a zụrụ azụ ruo ọgwụgwụ, mana na-agbakwunye ọtụtụ ndozi. Ọ na-eji koodu mgbanwe convolutional nwere ike nkwanye, ọtụtụ spectrogram na mfu nrụzi ngalaba oge, yana ndị mmegide maka ịdị mma nghọta. Ihe enyemaka pụtara ìhè bụ obere ihe atụ entropy dabere na Transformer nke na-emekọkwa koodu ndị ahụ na-enweghị atụ, na-apịpụta ntakịrị ntakịrị na-enweghị ezigbo mfu. EnCodec na-ewebatakwa ihe ngbanwe nke na-eme ka ọtụtụ mfu ọzụzụ na-asọmpi na-akpaghị aka ka ha wee kwụsie ike. Ọ na-ejikwa 24 kHz monophonic na 48 kHz stereo audio, na-arụ ọrụ gafee bitrates dị ka 1.5, 3, 6, na 12 kbps, na na 6 kbps rute àgwà tụnyere MP3 na 64 kbps. Ike akara ya Meta's MusicGen na AudioGen.
Nghọta nka nka
EnCodec's encoder wedara ihe atụ nke ebili mmiri na mgbagha gbagoro n'usoro ka ọ bụrụ usoro zoro ezo, nke RVQ na-atụgharị ka ọ bụrụ ndenye codebook. Ụdị asụsụ Transformer dị fechaa na-ebu amụma ihe gbasara puru omume na mgbakọ na mwepụ-koodu ha, na-enwetaghachi mkpakọ ọzọ n'efu. Onye na-ahụ maka nhazi ọzụzụ na-eweghachite onyinye gradient site na nrụzigharị, ihe nhụsianya, na mfu ndị na-emegide ya ka ọ nweghị otu okwu na-achịkwa, nke na-eme ka ọzụzụ ọtụtụ ebumnuche kwụsie ike n'ofe oke bitrate zuru oke.
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 EnCodec Audio mkpakọ
EnCodec abụrụlarị ihe nrịbama nke ndabara maka ọtụtụ ụdị ọdịyo na-emepe emepe, ụmụ ya na-akwalite ntụkwasị obi dị elu na obere bitrates, stereo zuru oke na nhazi ọkwa egwu, yana njikọta siri ike na ederede na ọdịyo na ndị na-emepụta egwu. Na-atụ anya nkuchi sara mbara na nkwurịta okwu bandwidth dị ala, mgbasa ozi ozugbo, yana dịka oyi akwa 'audio token' nke na-eme ka nnukwu ihe nlere asụsụ dị ka agụ ma dee ụda.
Mmejuputa n'ezie n'ụwa
Na-atụnye ụda olu maka Meta's MusicGen na AudioGen ndị na-emepụta ihe ederede gaa na ọdịyo
Na-akpakọ okwu 24 kHz na 1.5-6 kbps maka nnyefe nwere oke bandwidth
Na-ezobe egwu stereo 48 kHz nwere ogo dị nso MP3 na oke bitrates dị elu
Na-eje ozi dị ka koodu ntinye ebe mepere emepe maka nyocha na pipeline ML ọdịyo site na ebe nlele ewepụtara
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 EnCodec Audio Compression 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
Mgbakwunye ọdịyo na mmụta nnọchite anya
Ajụjụ a na-ajụkarị
What is EnCodec Audio Compression?
EnCodec bụ Meta's codec ụda olu nwere ntụkwasị obi dị elu nke na-akpakọ okwu na egwu n'ụdị bitrate dị ala nke nwere ụdị njiri mara mma. Ọ dị mkpa n'ihi na ọ na-akwado sistemụ ọdịyo ọgbara ọhụrụ yana ụgbọ mmiri n'ụdị mepere emepe maka onye ọ bụla iji.
Kedu otu wepụtara EnCodec?
Meta AI (FAIR) webatara EnCodec na 2022 dị ka codec ọdịyo akwara dị elu.
Kedu ihe mgbakwunye EnCodec na-agbakwunye na-enweghị mgbaghara karịa na akara ya?
EnCodec na-azụ obere Transformer ka ọ buru amụma gbasara puru omume na mgbakọ na mwepụ-koodu ha, na-enweta mkpakọ ọzọ na-enweghị mgbaghara n'elu RVQ.
Na roughly 6 kbps, EnCodec si àgwà ka a kọrọ dị ka tụnyere MP3 na ihe bitrate kedu?
EnCodec na 6 kbps na-erute ọdịdị nke onwe ya na MP3 na 64 kbps, nnukwu uru arụmọrụ.
Kedu nsogbu EnCodec's 'balancer' na-edozi n'oge ọzụzụ?
N'ọtụtụ mfu (nrụgharị, spectral, adversarial), onye na-edozi ahụ na-emezigharị gradients ha ka ọ nweghị otu ebumnuche na-achịkwa, na-eme ka ọzụzụ kwụsie ike.
Kedu usoro ngụkọ ọnụ ka EnCodec na-eso SoundStream na-ekekọrịta?
Dị ka SoundStream, EnCodec na-eji ọnụọgụ vector fọdụrụnụ iji mebie ntinye ntinye n'ime indices codebook.