Rabuwar Hanyoyi Dual-Path RNN
Dual-Path RNN (DPRNN) gine-ginen rarrabuwar sauti ne wanda ke raba jerin fa'idodin sauti masu tsayi zuwa gajerun hanyoyi masu ruɓani da sarrafa su ta hanyoyi guda biyu don haka cibiyoyin sadarwa na yau da kullun za su iya yin ƙira ga cikakkun bayanai na gida da tsarin duniya.
Dubawa
It matters because it made high-quality separation of long recordings practical.
Zurfafa nutsewa
Cibiyoyin sadarwa na yau da kullun suna gwagwarmaya tare da jeri mai tsayi sosai, kuma sautin yanki na lokaci a ƙimar samfura mai girma yana samar da jeri tare da dubun dubatar matakai. DPRNN (2020, Luo, Chen, Yoshioka) yana magance wannan ta hanyar sake fasalin fasalin zuwa grid na 2D na juzu'i. Sannan yana canza izinin wucewa na RNN guda biyu: ƙirar RNN mai intra-chunk na ɗan gajeren lokaci, ƙirar gida a cikin kowane ƙugiya, da ƙirar RNN mai tsaka-tsaki na dogon lokaci a cikin chunks. Tsara da yawa daga cikin waɗannan tubalan-hanyoyi biyu yana ba da damar ƙirar kama mahallin mahallin gabaɗayan furci yayin da kowane RNN ɗaya kawai yake ganin taga mai iya sarrafawa, ƙaramin jerin-tsayi. An jefar da shi cikin tsarin Conv-TasNet a matsayin wanda zai maye gurbin mai raba TCN, DPRNN ya ba da babbar riba a cikin ingancin rabuwa tare da ƙaramin adadin siga.
Fahimtar Fasaha
Makullin tsarin shine rarrabuwa tare da maimaita maimaitawa. Dogon jerin tsayi L yana naɗewa cikin matrix na K chunks na tsawon S (tare da 50% zoba). Intra-chunk RNN yana gudana tare da S (na gida), sannan tsaka-tsakin RNN yana gudana tare da K (duniya), kowannensu yawanci bidirectional. Saboda kowane tsarin RNN kawai matakan S ko K, haɓakawa yana tsayawa tsayin daka kuma ingantaccen filin karɓa ya zama cikakken jerin bayan ƴan tubalan. Haɓaka-ƙara yana sake gina jerin.
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 Rabuwar Hanyoyi Dual-Dual RNN
Tunanin hanya biyu na DPRNN ya zama samfuri wanda ya wuce takamaiman ƙwayoyin RNN ɗin sa. Babban nasara SepFormer ya musanya RNNs don Transformers a cikin tsarin intra/inter chunk iri ɗaya, kuma TF-GridNet ya tsawaita sarrafa-hanyoyi biyu cikin lokaci da mitar. Yi tsammanin tsarin rarrabuwa-da-madadin zai kasance madaidaicin tubalin ginin don ƙirar sauti mai tsayi, ƙara haɗe-haɗe tare da kulawa da amfani da bayan magana zuwa kiɗa da rarrabuwar sauti gabaɗaya.
Aiwatar da Gaskiyar Duniya
Rarraba masu magana da yawa a lokaci guda a cikin dogon taro ko rikodin hira.
Ƙarfafa ƙashin bayan intra/inter-chunk daga baya wanda SepFormer ya daidaita don rabuwa na zamani.
Ware muryar da aka yi niyya don rubutawa a ƙasa a cikin hayaniya, tattaunawa mai ma'ana.
Tsaftace sauti mai tsawo kamar laccoci ko tattaunawa inda masu magana ke magana akan juna.
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 Dual-Path RNN Separation 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
Samfuran RNN-Transducer
Tambayoyin da ake yawan yi
What is Dual-Path RNN Separation?
Dual-Path RNN (DPRNN) gine-ginen rarrabuwar sauti ne wanda ke raba jerin fa'idodin sauti masu tsayi zuwa gajerun hanyoyi masu ruɓani da sarrafa su ta hanyoyi guda biyu don haka cibiyoyin sadarwa na yau da kullun za su iya yin ƙira ga cikakkun bayanai na gida da tsarin duniya. Yana da mahimmanci saboda ya sanya rarrabuwar ɗorewa na dogon rikodin aiki mai inganci.
Wace matsala Dual-Path RNN ke magance da farko?
DPRNN tana sake tsara jerin yanki na dogon lokaci zuwa gungu-gungu don haka RNNs za su iya sarrafa mahallin gida da na duniya baki ɗaya ba tare da shaƙa tsawon lokaci ba.
Menene 'hanyoyin' guda biyu a cikin RNN Dual-Path?
Tsarin RNN guda ɗaya a cikin kowane yanki don tsarin gida; da sauran matakai a fadin chunks don dogon zangon tsari.
Yaya aka shirya jerin dogon fasali kafin toshe hanyoyin biyu?
DPRNN tana ninke jerin dogayen jeri zuwa matrix na juzu'i masu ruɓani don haka kowace RNN tana aiwatar da gajeriyar taga.
A cikin wanne tsari ne aka jefa DPRNN a matsayin maye gurbin?
DPRNN ta maye gurbin mai raba TCN a cikin bututun Conv-TasNet, yana adana rikodin koyo da dikodi.
Wanne samfurin daga baya ya kiyaye tsarin DPRNN na biyu amma ya maye gurbin RNNs da Transformers?
SepFormer ya sake yin amfani da ƙirar intra/inter-chunk dual-that design amma ya musanya sel masu maimaitawa don mai jujjuyawar kai.