1.CVC uye DSP kuderedza ruzha:
Kana vatengi vachitenga mahedhifoni eBluetooth, vanozogara vachinzwa CVC uye DSP kudzikisa ruzha mabasa ane vatengesi mukusimudzira mahedhifoni. Hazvina mhosva kuti vangani vashandisi vakanzwa tsananguro, vatengi vazhinji havasati vanzwisisa musiyano uripo pakati pezviviri izvi. Musiyano, wedambudziko rakadaro rehunyanzvi, tinouya kune sainzi yevaviri pasi pemutemo wekushanda uye musiyano.
DSP ipfupi yedhijitari masaini kugadzirisa. Nheyo yekushanda kwayo: maikorofoni inounganidza ruzha rwekunze rwezvakatipoteredza, uyezve kuburikidza neruzha rwekuderedza hurongwa hwekuita mukati memunzeve, inodzokorora kugadzira reverse sound wave yakaenzana neruzha rwakapoteredza, iyo inodzima ruzha uye nokudaro inowana zvakawanda. Zvakanaka kuderedza ruzha.
CVC ipfupi yeClear Voice Capture. Iyo isoftware yekudzikisa ruzha tekinoroji. Nheyo iyi ndeyekudzvanya mhando dzakasiyana dzeruzha rwekudzoreredza kuburikidza neyakavakirwa-mukati ruzha kudzima software nemakrofoni.
Musiyano sezvinotevera:
a. nokuti chinhu chakasiyana, CVC tekinoroji inonyanya kuitirwa echo inogadzirwa panguva yekufona, DSP inonyanya kukwira uye yakaderera frequency ruzha munzvimbo yekunze.
b. vakasiyana vanobatsirwa, tekinoroji yeDSP inonyanya kuita kuti vashandisi vemusoro vawane mari yavo, uye CVC inonyanya kubatsira mumwe bato.
Muchidimbu, mahedhifoni anoshandisa DSP uye CVC kudzikisa ruzha tekinoroji anogona kunyatso kudzikisa ruzha rwekunze kwenzvimbo yekufona, uye zvakanyanya kunatsiridza kunaka kwekufona uye ruzha rwemahedhifoni.
2.ANC ruzha kuderedza:
ANC inoreva Active Noise Control, iyo inoshingaira kuderedza ruzha. Nheyo yakakosha ndeyekuti iyo yekudzikisa ruzha system inoburitsa masaisai eruzha akaenzana neruzha rwekunze, achidzikamisa ruzha. Mufananidzo 1 idhizaini yedhizaini yefeedforward inoshanda ruzha kudzima nzeve. Iyo ANC chip inoiswa mukati memunzeve. Ref mic (reference maikorofoni) inounganidza ambient ruzha pamaearphone. Kukanganisa maikorofoni (Kukanganisa Maikorofoni) Inounganidza yasara ruzha mushure mekuderedza ruzha munzeve. Mutauri anoridza anti-ruzha mushure mekugadzirisa ANC.
Mufananidzo 2 idhizaini yehurongwa hweANC, ine mitsara mitatu, yakaparadzaniswa nemitsetse yakadhindwa. Iyo yepamusoro-soro nzira ndeye acoustic chiteshi kubva ku ref mic kuenda kukanganisa mic, iyo yekupindura basa inomiririrwa neP (z) P (z); iyo yepakati dhizaini ndiyo analogi chiteshi, uko yechipiri nzira ndiyo nzira kubva kune adaptive sefa inobuda kune inodzoka yasara. Kusanganisira DAC, reconstruction filter, power amplifier, speaker playback, re-acquisition, pre-amplifier, anti-aliasing filter, ADC; iyo yepasi pasi ndiyo nzira yedhijitari, uko inogadzirisa sefa inogara ichigadzirisa uremu hwesefa kuti uderedze zvakasara kusvika pasangana. Mhinduro yakajairika ndeyekushandisa inogadzirisa sefa uchishandisa FIR sefa musanganiswa neLMS algorithm. Rerutsa Mufananidzo 2 uye tora Mufananidzo 3.
Rega nditaure muchidimbu nezvemisimboti yeadaptive sefa uye LMS (Isingarevi sikweya) algorithm, uye ipapo Mufananidzo 3. Sezvinoratidzwa mumufananidzo 4, yakapihwa yekuisa xx uye inodiwa inobuda dd, inogadzirisa sefa inogadziridza coefficients pese pese kuitira kuti. musiyano uripo pakati pezvinobuda yy nadd unova diki uye diki kusvika chinosara chave pedyo zvekusvika zero uye chasangana. LMS is an update algorithm ye adaptive filters. Chinangwa cheLMS ndicho chikwere chezvikanganiso zvepakarepo e2(n)=(d(n)−y(n))2e2(n)=(d(n)−y(n))2, kuti uderedze. basa rechinangwa, Kushandisa iyo gradient descent inopa yakagadziridzwa formula yealgorithm. (Pfungwa yealgorithmic yekushandisa gradient descent kuderedza chinangwa uye kuwana formula yakagadziridzwa yeparameta ichatsvagwa yakajairika, senge mutsara regression.) Iyo yekuvandudza formula yeLMS algorithm uchishandisa FIR sefa ndeiyi: w(n+1) ) =w(n)+μe(n)x(n)w(n+1)=w(n)+μe(n)x(n), apo μμ iri nhanho saizi. Kana iyo μμ saizi yakagadziriswa nekudzokorora, idanho-ne-nhanho LMS algorithm.
Ngatitaurei nezve Mufananidzo 3. Pano iyo adaptive sefa inobuda mushure meS(z)S(z) kuenzanisa nechishuwo chinobuda. S(z)S(z) ichakonzera kusagadzikana. Muzvinyorwa, "chiratidzo chekukanganisa hachina 'kurongeka' nenzira kwayo Nenguva nechiratidzo chechiratidzo", kuchinjika kweLMS kwakaputsika. (Handisati ndaziva zvazvinoreva T__T) Nzira inoshanda ndeiyi FXLMS (Filtered-X LMS), iyo inobvumira x(n) kuiswa kuLMS module kuburikidza neSˆ(z)S^(z), Sˆ( z S^(z) ifungidziro yeS(z)S(z).Chinangwa cheFXLMS:
E2(n)=(d(n)−s(n)∗[wT(n)x(n)])2,
E2(n)=(d(n)−s(n)∗[wT(n)x(n)])2,
Saka gradient=−2e(n)s(n)∗x(n)−2e(n)s(n)∗x(n), uko s(n)s(n) isingazivikanwe, nefungidziro yaro, saka FXLMS Yekuvandudza fomula ndeye
w(n+1)=w(n)+μe(n)x'(n),
w(n+1)=w(n)+μe(n)x'(n),
Apo x'(n)=sˆ(n)∗x(n)x'(n)=s^(n)∗x(n).
Kana iyo adaptive sefa yasangana, E(z)=X(z)P(z)−X(z)W(z)S(z)≈0E(z)=X(z)P(z)−X(z ) W(z)S(z) ≈ 0, saka W(z) ≈ P(z) / S(z) W(z) ≈ P(z) / S(z). Kureva kuti, huremu hwehuremu hweiyo adaptive sefa inotarwa neiyo nzira yekutanga uye yechipiri nzira yehedhifoni. Iyo yekutanga nzira uye yechipiri nzira yehedhiseti yakagadzikana, saka huremu coefficient yeadaptive sefa yakagadzikana. Naizvozvo, nekuda kwekureruka, huremu huremu hwevamwe vanogadzira 'ANC mahedhifoni anotemerwa pafekitori. Ehe, chiitiko chekuteerera cheiyi ANC earphone zviri pachena kuti haina kunaka senge ANC earphone ine yechokwadi inoshanduka chirevo, nekuti mumamiriro ezvinhu chaiwo, ruzha rwekunze runoenderana negwara renzeve, tembiricha yakasiyana uye nezvimwe zvingave zvine pesvedzero. mhinduro yechiteshi cheearphone.
Matlab verification
Nyora Matlab kodhi, uchishandisa iyo inogadzirisa sefa yeinochinja nhanho saizi LMS, mhedzisiro yekuenzanisa inoratidzwa muMufananidzo 5. Muchikamu che 0 kusvika 2 kHz, iyo feedforward ANC inoshandiswa kubvisa Gaussian chena ruzha, uye ruzha attenuation i30 dB + paavhareji. Iyo FXLMS muMatlab raibhurari yakagadziriswa-nhanho, uye mhedzisiro yacho yakaipisisa.
Q&A
a. Nei ANC ichingori yeruzha rwakaderera pazasi 2 kHz?
Kune rimwe divi, iyo yekuvharisa ruzha rwemahedhifoni (passive ruzha kuderedzwa) inogona kunyatso kuvharira yakakwira-frequency ruzha, uye hazvifanirwe kushandisa ANC kuderedza yakakwirira-frequency ruzha. Nekune rimwe divi, ruzha rwepasi-rune ruzha rurefu uye runogona kutsungirira kumwe kunonoka kwechikamu, nepo ruzha rwepamusoro-soro rune hupfupi hwewavelength uye rwunonzwa nekutsauka kwechikamu, saka ANC inobvisa ruzha rwepamusoro.
b. Kana kunonoka kwemagetsi kwakakura kupfuura kunonoka kwekutanga, kuita kwegorgorithm kunogona kuderedzwa zvakanyanya?
P(z) kunonoka kudiki, S(z) kunonoka kwakakura, sekuti P(z)=z-1, S(z)=z-2, chete kana W(z)=z ichikwanisa kuita zvinodiwa, kwete. -kukonzera, Kusasvikika.
c. Ndeupi musiyano uripo pakati peFeedforward ANC, nhete-bhendi feedforward ANC, uye mhinduro ANC?
Iyo Feedforwad chimiro ine ref mic uye kukanganisa mic inounganidza kunze ruzha uye mukati masara masaini masaini, zvichiteerana. Chimiro chemhinduro chinongova nemhosho imwe chete mic, uye chiratidzo chereferensi chinogadzirwa nemhosho mic uye inogadzirisa sefa kubuda.
Iyo Broad-band feedforward ndiyo chimiro chakatsanangurwa pamusoro. Mune yakamanikana-bhendi chimiro, iyo ruzha sosi inogadzira chiratidzo chinokonzeresa chiratidzo jenareta, uye jenareta yechiratidzo inogadzira chiratidzo chechiratidzo cheiyo inogadzirisa sefa. Inoshanda chete pakubvisa periodic ruzha.
Feedback ANC inoshandisa kukanganisa mic kudzoreredza chiratidzo chakaunganidzwa neref mic mune feedforward chimiro nekuti inongova nemhosho mic. Iyo nzira haigutse iyo causal constraining, saka chete zvinofanotaurwa zveruzha zvikamu, kureva narrowband periodic ruzha, zvinobviswa. Izvo zvinofanirwa kucherechedzwa kuti kana iyo feedforward isingagutse iyo causal constraining, kureva kunonoka kwemagetsi kwakareba kupfuura iyo main channel acoustic kunonoka, inogona chete kubvisa narrowband periodic ruzha.
Kune zvakare Hybrid ANC chimiro chinosanganisira ese ari maviri feedforward uye mhinduro zvimiro. Kubatsira kukuru ndeyekuti iwe unogona kuchengetedza kurongeka kweiyo adaptive sefa.
