Ọnụọgụ Cepstral nke Mel-Frequency
Mel-Frequency Cepstral Coefficients (MFCC) bụ ọnụ ọgụgụ kọmpat nke na-achịkọta udi ụda ugboro ugboro ka ntị mmadụ si aghọta ya.
Nchịkọta
For decades they were the workhorse feature for speech recognition, speaker identification, and music analysis.
Ime miri emi
MFCC na-atụgharị obere iberi ọdịyo ka ọ bụrụ ọnụọgụ 13 na-ejide timbre ya. Ọkpụkpọ ahụ na-ewere ụdị ebili mmiri ahụ, gbajie ya n'ime okpokolo agba ~ 25ms, na-agbakọ ụdị ike dị iche iche site na mgbanwe Fourier, wee na-atụgharị axis ugboro ugboro na nha mel, nke na-ekesa otu ụzọ cochlea si eme: nke ọma n'okpuru 1kHz na nke dị n'elu. Ike mel bụ log-compressed (na-eṅomi nghọta ụda) ma mesịa gafere ngbanwe cosine dị iche, nke na-achọ ha mma ma na-etinye ozi n'ime ọnụọgụ ole na ole mbụ. Nsonaazụ siri ike na mkpọtụ na ụda okwu, nke mere na usoro okwu okwu nzuzo zoro ezo Markov Model na Gaussian Mixture Model ji dabere na MMFC ihe fọrọ nke nta ka ọ bụrụ ụwa niile tupu mmụta miri emi.
Nghọta nka nka
Ọnụ ọgụgụ mel na-eme ka nghọta dị nso na mel = 2595 log10 (1 + f/700), ya mere, nha nha nha nha nha nha nhata. The ikpeazụ discrete cosine mgbanwe (DCT) bụ 'cepstral' nzọụkwụ: ọ na-emeso log-mel ụdịdị dị ka ihe mgbaàmà na-ekewapụ nwayọọ nwayọọ na-agbanwe ụda olu-traktịp udi (obere cepstral coefficients, akụkụ anyị na-edobe) si ngwa ngwa pitch harmonics (elu coefficients, na-emekarị tụfuo), nke ọma na-ekewapụ njirimara phonetic site na ụda ọkà okwu.
Mmetụta atụmatụ
Nweta na iru
Ọ na-eme ka nnweta ya dịkwuo mma site na ndegharị, ịkọ akụkọ, na ntụgharị olu.
Ọnụ ego na mmefu ego
Ndị otu mgbasa ozi nwere ike ibubata ọdịyo a na-egbu maramara ngwa ngwa site na iji obere mmefu ego.
Ọsọ na ọnụ ọgụgụ
Sistemụ na-eche ihu ndị ahịa nwere ike hazie mkparịta ụka n'ọtụtụ buru ibu.
Ọdịnihu nke ọnụọgụ Cepstral nke Mel-frequency
Netwọk dị omimi nke ngwụcha ruo na ngwụcha na-amụtakwu atụmatụ ozugbo site na ụdị waveforms raw ma ọ bụ log-mel spectrograms, na-amafe DCT, yabụ MVCC dị ọcha na-apụ n'anya site na ASR ọgbara ọhụrụ. N'agbanyeghị nke ahụ, ha ka na-ewu ewu maka ọrụ dị arọ, ngwaọrụ na obere data: nchọpụta isiokwu, nchọpụta ọrụ olu, mkpisi aka ọdịyo, na bioacoustics. Na-atụ anya ka MMFCs nọgide na-adị ka usoro nhazi nke ọma, nke a pụrụ ịkọwapụta ọbụna dị ka njedebe n'ihu amụtara na-achịkwa nnukwu ụdị.
Mmejuputa n'ezie n'ụwa
Njirimara Acoustic maka ndị amata okwu HMM-GMM dị ka usoro Sphinx mbụ na HTK
Nnwale okwu na nkwuputa okwu, ịmata onye na-ekwu okwu na oku
Nhazi ụdị egwu na akara mkpịsị aka egwu (Timbre nke ụdị Shazam)
Ịchọta mmejọ igwe ma ọ bụ oku anụmanụ sitere na ọdịyo na ụlọ ọrụ mmepụta ihe na nlekota bioacoustic
Ihe ize ndụ & okporo ụzọ nche
Iji olu eme ihe na ihe egwu mpụta ga-abawanye mgbe nkwenye na-efu.
Izi ezi nwere ike ịdaba n'ofe ụda olu, olumba ma ọ bụ gburugburu mkpọtụ.
Enwere ike imehie ọdịyo sịntetik dị ka ezigbo okwu na-enweghị akara doro anya.
Map mmejuputa
Nweta nkwenye doro anya maka ijide olu, imechi, na ijigharị.
Nwale ogo n'ofe ndị na-ekwu okwu dị iche iche yana ọnọdụ ndabere.
Kọwaa mgbe mmadụ ga-enyocha ma ọ bụ kwado nsonye.
Deba aha ọdịyo sịntetik ma debe ndekọ ihe ndekọ maka ịza ajụjụ.
Nọgide na-eme nchọpụta
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 Mel-Frequency Cepstral Coefficients 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
Ntuziaka na-esote
Mel Spectrogram
Ajụjụ a na-ajụkarị
What is Mel-Frequency Cepstral Coefficients?
Mel-Frequency Cepstral Coefficients (MFCC) bụ ọnụ ọgụgụ kọmpat nke na-achịkọta udi ụda ugboro ugboro ka ntị mmadụ si aghọta ya. Ruo ọtụtụ iri afọ, ha bụ njirimara ọrụ maka njirimara okwu, njirimara ọkà okwu, na nyocha egwu.
Kedu ihe 'mel' na MCCC na-ezo aka?
Ọnụ ọgụgụ mel na-adaba ugboro ole ka nha nha nha na-ada ụda n'ebe dị anya na ụmụ mmadụ, gbasasịrị nke ọma na obere ugboro na nke dị elu.
Kedu mgbanwe a na-etinye n'ikpeazụ iji mepụta ọnụọgụ cepsral?
Mgbe agbakọchara ike log-mel, mgbanwe cosine dị iche na-emezi ha mma ma na-ebukọta ozi n'ime ọnụọgụ ole na ole mbụ.
Kedu ihe kpatara MFCC ji sie ike na ọkwa ọkà okwu?
Ọnụọgụ cepstral dị ala na-ejide envelopu ụda olu (ọdịnaya ụda olu) ebe ndị dị elu na-ejide ụda, ya mere idobe ndị dị ala na-emesi ụda ike ike.
Ọ dị ka ọnụọgụ ọnụọgụ MCCC ole ka edobere n'otu etiti?
Nhọrọ a na-ahụkarị bụ ihe dị ka ọnụọgụ 13, mgbe ụfọdụ a na-agbakwunye ya na delta ha, nke na-achịkọta timbre nke ọma.
Kedu ihe kpatara eji etinye log ahụ na ike mel-band?
Nghọta ụda olu mmadụ bụ ihe dị ka logarithmic, yabụ mkpakọ log na-eme ka njirimara ndị ahụ dịkwuo mma ma na-eme ka ọ dị ike.