Kayan aikin Gane Magana Kaldi
Kaldi kyauta ce, buɗaɗɗen kayan aikin kayan aiki wanda ya zama babban dandalin bincike don gina tsarin tantance magana.
Dubawa
It matters because for nearly a decade it was the go-to foundation for academic and industrial ASR work.
Zurfafa nutsewa
Kaldi, wanda aka saki a cikin 2011 kuma Daniel Povey ya jagoranta, an rubuta shi a cikin C ++ tare da girke-girke da aka haɗa tare da rubutun bash da Perl. An gina shi akan bututun ASR na yau da kullun: cire fasalin sauti (MFCCs ko filterbanks), ƙirar ƙirar sauti tare da Gaussian Mixture Models ko, daga baya, hanyoyin sadarwa mai zurfi, da haɗa ƙirar sauti, ƙamus na furci, da ƙirar harshe cikin jadawali guda ɗaya da ake iya nema. Ma'anar zaɓinta na fasaha shine ta amfani da ma'auni masu ƙayyadaddun iyaka (WFSTs) daga ɗakin karatu na OpenFST don tsara duk tushen ilimi cikin jadawali guda ɗaya. Kaldi ya aika da 'kayan girke-girke' don daidaitattun saitunan bayanai kamar Switchboard, Librispeech, da Wall Street Journal, barin masu bincike su sake fitar da sakamako na zamani. Ya zama aiwatar da tunani wanda aka yiwa sabbin tsarin ƙima.
Fahimtar Fasaha
Babban dabarar Kaldi shine hada WFSTs guda hudu cikin jadawali daya da ake kira HCLG: H taswirorin neural-net ko jihohin GMM zuwa wayoyi masu dogaro da mahallin, C tana sarrafa mahallin sauti (wayoyin triphones), L shine ƙamus na taswirar wayoyi zuwa kalmomi, kuma G shine ƙirar harshe. Ƙirƙirar waɗannan masu fassara da haɓaka sakamakon yana samar da jadawali ɗaya mai ƙididdigewa yana bincika tare da tsinkewar Viterbi algorithm, yana mai da firam ɗin sauti cikin mafi kusantar jerin kalmomi da inganci.
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 Kayan Aikin Gane Maganar Kaldi
Tsarin HMM-DNN na Kaldi an maye gurbinsa da ƙirar jijiyoyi na ƙarshe zuwa ƙarshe waɗanda ke taswirar sauti kai tsaye zuwa rubutu. Aikin magaji na Daniel Povey, k2 (tare da yanayin yanayin Icefall da Lhotse), yana sake tunanin ra'ayoyin WFST na Kaldi a cikin PyTorch tare da rarrabuwa ta atomatik-jihar automata. Yi tsammanin Kaldi da kansa ya zama abin tuntuɓar tarihi da kayan aikin koyarwa, yayin da zuriyarsa ta ra'ayi ta haɗu da tsararren tsari na gargajiya tare da tsarin taswira na zamani da ƙirar sauti mai kulawa da kai.
Aiwatar da Gaskiyar Duniya
Dakunan gwaje-gwaje na ilimi suna sake haifar da Librispeech da ma'auni na Switchboard don inganta sabon binciken ƙirar ƙira.
Gina tsarin umarnin murya na al'ada don ƙananan albarkatu ko ƙananan harsuna ta amfani da girke-girke na Kaldi
Tilastawa mai jiwuwa zuwa rubuce-rubuce don ilimin harshe, ƙirƙirar saitin bayanai, da lokacin juzu'i
Ƙaddamar da binciken murya da wuri da ƙaƙƙarfan ƙaƙƙarfan baya a cikin masana'antu kafin ƙirar ƙarshe-zuwa-ƙarshen balaga
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 Kaldi Speech Recognition Toolkit 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
Gane Jawabin Wawasi
Tambayoyin da ake yawan yi
What is Kaldi Speech Recognition Toolkit?
Kaldi kyauta ce, buɗaɗɗen kayan aikin kayan aiki wanda ya zama babban dandalin bincike don gina tsarin tantance magana. Yana da mahimmanci saboda kusan shekaru goma shine tafi-zuwa tushe don aikin ASR na ilimi da masana'antu.
Wanne yaren shirye-shirye aka rubuta ainihin Kaldi?
An rubuta ainihin Kaldi a cikin C++ don yin aiki, tare da rubutun bash da Perl waɗanda ke tsara horo da yanke girke-girke.
Wane tsarin lissafi Kaldi yake amfani da shi don haɗa samfurin sauti, ƙamus, da ƙirar harshe zuwa jadawali guda ɗaya?
Kaldi yana tsara WFSTs daga ɗakin karatu na OpenFST zuwa cikin jadawali ɗaya na HCLG wanda mai yankewa yake nema.
Wanene farkon mahalicci kuma jagoran aikin Kaldi?
Daniel Povey ya jagoranci ci gaban Kaldi, wanda aka fara bugawa a cikin 2011, sannan ya fara aikin magaji k2.
A cikin bututun gargajiya na Kaldi, menene GMMs (Gaussian Mixture Model) waɗanda aka fara amfani da su don yin ƙira?
GMMs sun ƙirƙira da yuwuwar sautin sauti na jihohin sauti kafin zurfafa hanyoyin sadarwar jijiyoyi su maye gurbinsu a cikin tsarin gauraye.
Menene sunan tsarin muhalli na tushen PyTorch na zamani ga Kaldi?
k2, tare da Icefall da Lhotse, suna sake aiwatar da ƙayyadaddun ra'ayoyin Kaldi a cikin wani nau'i na abokantaka na PyTorch.