SoundStorm Parallel Audio Generation
SoundStorm samfurin tsarar sauti Google ne wanda ke samar da magana da sauti a layi daya maimakon alamar guda ɗaya a lokaci guda, yana samar da ingantaccen sauti mai inganci cikin sauri.
Dubawa
It matters because it cuts generation latency for long clips from minutes to seconds without sacrificing fidelity.
Zurfafa nutsewa
SoundStorm, wanda Google ya gabatar a cikin 2023, yana haifar da sauti mai wakilta a matsayin saƙon sauti mai hankali daga lambar kodi na jijiya mai suna SoundStream. Samfuran da suka gabata kamar AudioLM sun samar da waɗannan alamomin kai tsaye, suna tsinkayar kowane alamar a jere, wanda ke jinkirin dogon sauti. SoundStorm a maimakon haka yana amfani da hanyar da ba ta kai tsaye ba, tsarin tushen abin rufe fuska wanda aka aro daga ƙirar tsara hoto kamar MaskGIT. Yana farawa da galibin alamun rufe fuska kuma yana cika su sama da ɗimbin matakai na yanke hukunci, yana tsinkayar alamu da yawa lokaci guda a layi daya. Yanayi akan alamomin ma'ana (daga samfuri kamar AudioLM ko SPEAR-TTS), yana iya haɗa daƙiƙa 30 na tattaunawa ta halitta a cikin kusan rabin daƙiƙa akan TPU, kusan sau 100 cikin sauri fiye da layin autoregressive yayin dacewa da ingancinsu da daidaiton magana.
Fahimtar Fasaha
SoundStorm yana ƙirƙira matsayi na matakan ƙididdigewa na saura (RVQ) daga SoundStream. Yayin horo, ana rufe alamun bazuwar kuma samfurin ya koyi tsinkaya su. A cikin ƙididdiga yana gudanar da yanke hukunci daidai gwargwado na tushen aminci: a cikin kowane juzu'i yana tsinkayar duk abin rufe fuska, yana kiyaye mafi ƙarfin gwiwa, kuma yana sake rufe sauran. Yana ƙaddamar manyan matakan RVQ da farko, sannan mafi kyawu, yana kaiwa ga cikakken sauti cikin ƴan matakai kaɗan fiye da tsarar alama ta alama.
Dabarun Tasiri
Shiga ku isa
Yana inganta samun dama ta hanyar rubutu, ba da labari, da mu'amalar murya.
Kudin da kasafin kuɗi
Ƙungiyoyin kafofin watsa labaru na iya jigilar sauti mai gogewa cikin sauri tare da ƙaramin kasafin kuɗi.
Gudu da sikelin
Tsarin fuskantar abokin ciniki na iya aiwatar da hulɗar magana a mafi girman ma'auni.
Makomar SoundStorm Parallel Audio Generation
Daidaita tushen abin rufe fuska yana zama daidaitaccen kayan aiki don sauri, mai sarrafa sauti. Yi tsammanin zai ba da ikon wakilai na tattaunawa na ainihi, haɗin murya nan take, da kuma faifan bidiyo na dogon lokaci ko tsarar littattafan mai jiwuwa inda latency ya taɓa yin amfani da ƙira na autoregressive. Haɗa shi da ƙaƙƙarfan kwandishan na ma'ana da alamar ruwa zai inganta gaskiyar tattaunawa da ganowa. Irin wannan ra'ayin gyaran-gyare-gyare yana yiwuwa ya haɗu tare da hanyoyin watsawa, yana ɓatar da layi tsakanin codec-token da ci gaba da janareta na sauti.
Aiwatar da Gaskiyar Duniya
Samar da maganganun magana na daƙiƙa 30 don mataimakan muryar AI a ƙarƙashin daƙiƙa guda
Haɗa tattaunawa mai juyi da yawa tare da daidaitattun muryoyin lasifika don yin samfuri
Ƙaddamar da ƙaramin-latency rubutu-zuwa-magana a cikin ma'aikatu masu mu'amala inda samfuran autoregressive suka yi rauni
Ƙirƙirar sauti mai tsayi mai tsayi da sauri ta hanyar cike alamun sauti a layi daya
Hatsari & Tsare-tsare
Rashin amfani da murya da haɗarin kwaikwaya yana ƙaruwa lokacin da aka rasa izini.
Daidaituwa na iya faɗuwa cikin lafuzza, yaruka, ko mahalli masu hayaniya.
Ana iya kuskuren sauti na roba don ingantacciyar magana ba tare da bayyananniyar lakabi ba.
Taswirar Hanya
Sami tabbataccen izini don ɗaukar murya, cloning, da sake amfani.
Gwajin ingantattun masu magana daban-daban da yanayin baya.
Ƙayyade lokacin da dole ne ɗan adam ya duba ko ya amince da abubuwan da aka fitar.
Yi lakabin sauti na roba da kuma adana bayanan da aka tabbatar don yin lissafi.
Ci gaba da Bincike
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 SoundStorm Parallel Audio Generation 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
Jagora na gaba
Stable Audio Latent Diffusion
Tambayoyin da ake yawan yi
What is SoundStorm Parallel Audio Generation?
SoundStorm samfurin tsarar sauti Google ne wanda ke samar da magana da sauti a layi daya maimakon alamar guda ɗaya a lokaci guda, yana samar da ingantaccen sauti mai inganci cikin sauri. Yana da mahimmanci saboda yana yanke jinkirin tsara don dogon shirye-shiryen bidiyo daga mintuna zuwa daƙiƙa ba tare da sadaukar da aminci ba.
Menene mabuɗin ƙirƙira wanda ke sa SoundStorm sauri fiye da AudioLM?
SoundStorm yana maye gurbin jinkirin autoregressive, tsarar alama-by-token tare da yanke hukunci mara-autoregressive, yana tsinkayar alamu da yawa a lokaci guda.
Wace dabara daga tsarar hoto ce SoundStorm ke daidaitawa?
SoundStorm yana ɗaukar tushen dogaro, abin rufe fuska da cika dabarun yanke hukunci wanda MaskGIT ya shahara a cikin haɗin hoto.
Wane irin wakilci ne SoundStorm ke samarwa?
SoundStorm yana annabta sahihan alamun sautin sauti da aka samar ta hanyar codec na SoundStream, waɗanda daga baya aka canza su zuwa tsarin motsi.
Kusan yaushe SoundStorm ke ɗauka don samar da daƙiƙa 30 na sauti akan TPU?
SoundStorm na iya haɗa daƙiƙa 30 na sauti a cikin kusan daƙiƙa 0.5, kusan 100x cikin sauri fiye da madaidaicin tushe.
Ta yaya SoundStorm ke yanke shawarar waɗanne alamun annabta don kiyayewa yayin yanke hukunci?
A cikin kowane juzu'i yana riƙe da tsinkaya mafi girman ƙarfin ƙarfinta kuma yana sake rufe waɗanda ba su da tabbas don wucewa na gaba.