TWS headset isignali yedijithali ADM
Ngokukhula okuqhubekayo kwe-TWS (i-Tereo ye-wireless stereo) ye-headset market.Iimfuno zabasebenzisi zamava emveliso nazo ziphuculwe ukusuka kwikhonkco elikhawulezayo ukuya kwimigangatho ephezulu.Ngokomzekelo, ukususela kulo nyaka, inani elikhulu lee-headset ze-TWS ezineefowuni ezicacileyo ziye zavela kwimarike.
Ukwenza unxibelelwano lwezwi olucacileyo kwiindawo ezinomsindo kakhulu, ngaba kunokwenzeka ukuvelisa izikimu ezidibanisa izibonakaliso ezivela kwindlebe yangaphakathi kunye neemakrofoni zangaphandle ukuphumeza iteknoloji ehlakaniphile, ehambelana ne-sub-band mixer technology.Enyanisweni, ezinye iinkampani ze-algorithm zasekhaya kunye nezangaphandle zizinikele kule nto, kwaye ziphumelele iziphumo ezithile.
Ngokuqinisekileyo, iinkampani ezininzi zesisombululo ngoku zigxininise ngokukodwa kwizisombululo zokunciphisa ingxolo yomnxeba njenge-AI edge (le enye), kodwa eneneni, iphuculwe ngakumbi kwizisombululo zokunciphisa ingxolo ekhoyo, ngoko le nxalenye isusiwe, makhe sijonge. amanye amalungu asisiseko kuqala Intshayelelo, oko kukuthi, yintoni enokwenziwa yifowuni yokunciphisa ingxolo.
Ngokubanzi, ukufowuna ukunciphisa ingxolo kuxhomekeke kwi-Uplink (uplink) kunye ne-Downlink (downlink) ukuvumelanisa.Ngokumalunga neMicrophone Array/AEC/NS/EQ/AGC/DRC, ubudlelwane obunengqiqo bumi ngolu hlobo lulandelayo:
I-ADM (i-Adaptive Directional Microphone Array) iteknoloji yokucwangcisa i-digital signal eyenza i-microphone ye-directional okanye i-ngxolo-cancelling microphone usebenzisa i-microphone ezimbini ze-omnidirectional kuphela.I-ADM itshintsha ngokuzenzekelayo iimpawu zayo zokuqondisa ukubonelela ngokunciphisa ingxolo kwiindawo ezahlukeneyo ngelixa igcina umgangatho womqondiso owaneleyo.Inkqubo yokulungelelanisa ikhawuleza, inokukhetha okunamandla okuphindaphindayo, kwaye inokuphelisa ukuphazamiseka okuninzi ngaxeshanye.
Ukongeza kwiimpawu zayo ezilungileyo zokwalatha, ii-ADM zichaphazeleka ngakumbi kwingxolo yomoya kunemibhobho yesiqhelo ye-acoustic.Itekhnoloji ye-ADM ivumela iintlobo ezimbini zokucwangciswa kwe-microphone: "endfire" kunye ne "broadfire".
Kwi-endfire configuration, umthombo wesignali (umlomo womsebenzisi) ukwi-axis (umgca odibanisa ii-microphone ezimbini).Kuqwalaselo olubanzi, lujolise kumgca othe ngqo kwi-axis ethe tye.
Kwi-endfire configuration, i-ADM ineendlela ezimbini zokusebenza;"Far-talk" kunye "close-talk".Kwimowudi ekude, i-ADM isebenza njenge-microphone yolwalathiso olufanelekileyo, igxininise isignali ukusuka ngasemva kunye namacala ngelixa igcina umqondiso ukusuka ngaphambili.Kwimo yokuthetha ngokusondeleyo, i-ADM isebenza njengeyona mibhobho ye-microphone yokucima ingxolo, ngokufanelekileyo ukuphelisa izandi ezikude.Inkululeko ehambelanayo yoyilo lwe-acoustic yenza ii-ADM zilungele iiselfowuni, ezivumela ukuba "i-soft" itshintshe phakathi kwezithethi ezikude kunye nezithethi ezikufutshane.Nangona kunjalo, xa olu hlobo loyilo lusetyenziswa kwii-earphones, ngakumbi ii-earphone ze-TWS, kuthintelwe ngakumbi ukuba umsebenzisi uyinxibe ngokuchanekileyo.Ngokufana nee-airpods, umbhali uye waqaphela ukuba abantu abaninzi "baneendlela ezingaqhelekanga" zokunxiba kwindlela engaphantsi, ezinye zazo ziindlebe zabasebenzisi.Ubume, kunye nemikhwa ethile yokunxiba, ibangela ukuba i-algorithm ingasebenzi kwimeko efanelekileyo.
I-Acoustic Echo Canceller (AEC)
Xa inxalenye yesignali kwi-duplex (ngaxeshanye-indlela-mbini) unxibelelwano lubuyela kwisignali yomthombo, kuthiwa "i-echo".Kwi-analog yomgama omde kwaye phantse zonke iinkqubo zonxibelelwano lwedijithali, neempawu ezincinci ze-echo zinokubangela ukuphazamiseka ngenxa yolibaziseko olubi lokuya nokubuya.
Kwi-terminal yonxibelelwano lwezwi, i-acoustic echoes iveliswa ngenxa yokudibanisa okuvakalayo phakathi kwesithethi kunye nemakrofoni.Ngenxa yenkqubo engasebenziyo esetyenzisiweyo kumjelo wonxibelelwano, njengee-vocoder ezilahlekileyo kunye ne-transcoding, i-acoustic echoes kufuneka iqhutywe kwindawo (icinywe) ngaphakathi kwesixhobo.
Isicinezeli Sengxolo (NS)
Itekhnoloji yokunciphisa ingxolo inciphisa ingxolo emileyo kunye neyexeshana kwimiqondiso yentetho yetshaneli enye, iphucula umlinganiselo wesignali ukuya kwingxolo, iphucula ukuvakala kwentetho, kwaye inciphisa ukudinwa kweendlebe.
Ngokuqinisekileyo, zininzi iindlela ezithile kule nxalenye, njenge-BF (Beamforming), okanye i-PF (I-Post filter) kunye nezinye iindlela zohlengahlengiso.Ngokubanzi, i-AEC, i-NS, i-BF, kunye ne-PF zizinto eziphambili zokunciphisa ingxolo yokufowuna.Kuyinyani ukuba umboneleli ngamnye wesisombululo se-algorithm uneenzuzo ezahlukeneyo kunye nokungonakali.
Kwinkqubo yonxibelelwano yezwi eqhelekileyo, inqanaba lomqondiso welizwi linokwahluka kakhulu ngenxa yomgama phakathi komsebenzisi kunye nemakrofoni, kwaye ngenxa yeempawu zejelo lonxibelelwano.
Uxinzelelo lweDynamic Range (DRC) yeyona ndlela ilula yokulinganisa amanqanaba ophawu.Uxinzelelo kunciphisa uluhlu oluguquguqukayo lomqondiso ngokunciphisa (ukucinezela) izahlulo zentetho eyomeleleyo ngelixa ugcina ngokwaneleyo izahlulo zentetho ezibuthathaka.Ke ngoko, umqondiso wonke unokwandiswa ngakumbi ukuze imiqondiso ebuthathaka ivakale ngcono.
Itekhnoloji ye-AGC inyusa ngokwedijithali ukufumana isignali (ukukhulisa) xa isignali yelizwi ibuthathaka, kwaye iyicinezele xa isignali yelizwi yomelele.Kwiindawo ezinengxolo, abantu badla ngokuthethela phezulu, kwaye oku ngokuzenzekelayo kusetha ingeniso yesitishi semakrofoni kwixabiso elincinci, ngaloo ndlela kuncitshiswe ingxolo yendawo ngelixa igcina ilizwi lomdla likwinqanaba elifanelekileyo.Kwakhona, kwindawo ezolileyo, abantu bathetha ngokuzolileyo ukuze amazwi abo anyuswe yi-algorithm ngaphandle kwengxolo eninzi.
Ixesha lokuposa: Jun-07-2022
