Audio AI JAGORA

Conv-TasNet Lokaci-Yankin Rabewa

Conv-TasNet cibiyar sadarwa ce ta jijiya wacce ke raba gauraye audio (kamar mutane biyu suna magana lokaci guda) ta yin aiki kai tsaye akan danyen sautin raƙuman sauti maimakon siffa.

2 min karatuAn sabunta ta ƙarshe

Dubawa

Yana da mahimmanci saboda ya saita sabon mashaya don ingancin rabuwar magana yayin da yake gudana cikin sauri don amfani na ainihi.

Zurfafa nutsewa

Tsarin rarrabuwa na al'ada yana canza sauti zuwa na'ura mai ba da hanya tsakanin hanyoyin sadarwa, raba mitoci, sannan jujjuya baya, wanda ya rasa bayanin lokaci kuma yana da inganci. Conv-TasNet (2019, Luo da Mesgarani) ya tsallake hakan gaba daya. Yana amfani da mai rikodin koyo (convolution 1D) don juyar da gajerun nau'ikan igiyoyin igiyar ruwa zuwa madaidaicin wakilci na ciki, hanyar sadarwar rabuwa wacce ke kimanta abin rufe fuska ga kowane mai magana, da na'urar dikodi mai koyo wacce ke sake gina kowane tsaftataccen tsarin igiyar ruwa. Mai raba tari ne na faɗuwar juzu'i na 1D da ake kira Cibiyar Sadarwar Juyin Juya Hali (TCN), wanda ke ɗaukar mahallin dogon zango ba tare da maimaituwa ba. An horar da shi tare da asarar SI-SNR-ma'auni-ma'auni da horarwa-sau-lu'u, ya zarce abin rufe fuska mai kyau, sakamakon da zarar an yi tunanin ya zama babba.

Fahimtar Fasaha

Babban dabarar ita ce maye gurbin madaidaiciyar Canjin Canji na ɗan gajeren lokaci tare da mai rikodin juzu'i na 1D da aka koya, don haka hanyar sadarwar ta sami wakilcin sauti wanda aka inganta don abin rufe fuska maimakon wanda aka ƙera don kallon ɗan adam. Mai raba TCN yana amfani da juzu'i masu ɗimbin yawa tare da abubuwan haɓaka haɓakawa da yawa, yana ba da babban filin karɓu yayin kasancewa cikakke daidai gwargwado. Masks suna ninka rufaffiyar fasalulluka-hikima, kuma jujjuyawar juzu'in da aka canjawa ita ke yanke kowane abin rufe fuska zuwa tsarin igiyar ruwa.

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 Conv-TasNet Lokaci-Yankin Rabewa

Conv-TasNet ya samar da dukan dangin samfuran yanki na lokaci. Masu nasara kamar DPRNN, SepFormer, da TF-GridNet sun tura ingancin rabuwa sosai, amma Conv-TasNet ya kasance mai ƙarfi, tushe mai nauyi kuma har yanzu ana tura shi akan na'urar inda lissafin ke da ƙarfi. Yi tsammanin ƙaramin ƙirar TCN ɗin sa don ci gaba da bayyana a cikin na'urorin sauraron ji, belun kunne, da taron tattaunawa na lokaci-lokaci, galibi ana distilled ko ƙididdige su don gudana tsakanin milliseconds akan guntun wayar hannu.

Aiwatar da Gaskiyar Duniya

Rarraba lasifika biyu masu karo da juna a cikin taron da aka yi rikodi ta yadda kowane za a iya rubuta shi cikin tsafta.

Haɓaka magana a cikin na'urorin kunne da na'urorin ji waɗanda ke keɓance mai magana da magana daga baya.

Kafin aiwatar da sautin tsakiyar kira mai hayaniya kafin ciyar da shi zuwa tantance magana ta atomatik.

Tsaftace tattaunawa mai cin karo da juna a cikin podcast ko fim bayan samarwa.

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

1

Sami tabbataccen izini don ɗaukar murya, cloning, da sake amfani.

2

Gwajin ingantattun masu magana daban-daban da yanayin baya.

3

Ƙayyade lokacin da dole ne ɗan adam ya duba ko ya amince da abubuwan da aka fitar.

4

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 Conv-TasNet Time-Domain Separation quiz

Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.

Fara tambayoyi

Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation

Jagora na gaba

Buɗe-Unmix Rabewar Kiɗa

Tambayoyin da ake yawan yi

Menene Rabuwar Domain Lokaci na Conv-TasNet?

Conv-TasNet cibiyar sadarwa ce ta jijiya wacce ke raba gauraye audio (kamar mutane biyu suna magana lokaci guda) ta yin aiki kai tsaye akan danyen sautin raƙuman sauti maimakon siffa. Yana da mahimmanci saboda ya saita sabon mashaya don ingancin rabuwar magana yayin da yake gudana cikin sauri don amfani na lokaci-lokaci.

Menene ma'anar fasalin Conv-TasNet idan aka kwatanta da tsarin rabuwa na baya?

Conv-TasNet yana maye gurbin kafaffen bututun STFT tare da mai rikodin rikodi/dikodi don haka yana raba sauti a cikin yanki na lokaci akan sigar ƙaƙƙarfan ƙaƙƙarfan igiyar ruwa.

Wace irin hanyar sadarwa Conv-TasNet ke amfani da ita azaman tsarin rabuwarta?

Mai raba TCN ce da aka gina daga faifan juzu'i na 1D, yana ba da babban filin karɓa yayin da ya kasance mai daidaitawa.

Me zai maye gurbin Canji na gajeriyar Lokaci na gargajiya a cikin Conv-TasNet?

Maimakon kafaffen STFT, ƙwararren 1D juyin juya halin yana ɓoye nau'ikan nau'ikan igiyoyin ruwa zuwa wakilcin da aka inganta don rufe fuska.

Wane aikin asara ne ke tsakiyar horar da Conv-TasNet?

An horar da Conv-TasNet tare da asarar SI-SNR haɗe tare da horarwar da ba ta dace ba don ɗaukar odar da ba a sani ba na rabe-raben lasifika.

Wane gagarumin sakamako Conv-TasNet ya cimma kan rabuwar magana?

Ta hanyar nisantar asarar lokaci na hanyoyin duban abubuwa, Conv-TasNet ya zarce madaidaicin ma'aunin abin rufe fuska na lokaci-lokaci.