{"id":53722,"date":"2019-12-08T00:00:00","date_gmt":"2019-12-07T21:00:00","guid":{"rendered":"https:\/\/prohoster.info\/blog\/blog_prohoster\/kompaniya-mozilla-predstavila-dvizhok-raspoznavaniya-rechi-deepspeech-0-6"},"modified":"2020-02-18T14:01:39","modified_gmt":"2020-02-18T11:01:39","slug":"kompaniya-mozilla-predstavila-dvizhok-raspoznavaniya-rechi-deepspeech-0-6","status":"publish","type":"post","link":"https:\/\/prohoster.info\/cs\/blog\/news\/kompaniya-mozilla-predstavila-dvizhok-raspoznavaniya-rechi-deepspeech-0-6","title":{"rendered":"Spole\u010dnost Mozilla p\u0159edstavila engine pro rozpozn\u00e1v\u00e1n\u00ed \u0159e\u010di DeepSpeech 0.6.","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"<p><noindex><a rel=\"nofollow\" href=\"https:\/\/hacks.mozilla.org\/2019\/12\/deepspeech-0-6-mozillas-speech-to-text-engine\/\">P\u0159edstaven\u00ed<\/a><\/noindex> vyd\u00e1n\u00ed rozpozn\u00e1vac\u00edho motoru vyvinut\u00e9ho spole\u010dnost\u00ed Mozilla <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/mozilla\/DeepSpeech\/releases\/v0.6.0\">DeepSpeech 0.6<\/a><\/noindex>, kter\u00fd implementuje shodnou architekturu rozpozn\u00e1v\u00e1n\u00ed \u0159e\u010di, <noindex><a rel=\"nofollow\" href=\"https:\/\/arxiv.org\/abs\/1412.5567\">navr\u017eenou<\/a><\/noindex> v\u011bdci ze spole\u010dnosti Baidu. Implementace je napsan\u00e1 v jazyce Python s vyu\u017eit\u00edm platformy strojov\u00e9ho u\u010den\u00ed TensorFlow a <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/mozilla\/DeepSpeech\">distribuov\u00e1no<\/a><\/noindex> je licencov\u00e1na pod volnou licenc\u00ed MPL 2.0. Podporov\u00e1na je pr\u00e1ce v Linuxu, Androidu, macOS a Windows. V\u00fdkon je dostate\u010dn\u00fd pro pou\u017eit\u00ed motoru na desk\u00e1ch LePotato, Raspberry Pi 3 a Raspberry Pi 4.<\/p>\n<p>Sou\u010d\u00e1st\u00ed jsou tak\u00e9 <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/mozilla\/DeepSpeech\/releases\/download\/v0.6.0\/deepspeech-0.6.0-models.tar.gz\">jsou<\/a><\/noindex> nau\u010den\u00e9 modely, <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/mozilla\/DeepSpeech\/releases\/download\/v0.6.0\/audio-0.6.0.tar.gz\">p\u0159\u00edklad\u016f<\/a><\/noindex> zvukov\u00e9 soubory a n\u00e1stroje pro rozpozn\u00e1v\u00e1n\u00ed z p\u0159\u00edkazov\u00e9ho \u0159\u00e1dku. Pro implementaci funkce rozpozn\u00e1v\u00e1n\u00ed \u0159e\u010di do sv\u00fdch program\u016f jsou k dispozici hotov\u00e9 moduly pro Python, NodeJS, C++ a .NET (extern\u00ed v\u00fdvoj\u00e1\u0159i p\u0159ipravili samostatn\u00e9 moduly pro <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/RustAudio\/deepspeech-rs\">Rust<\/a><\/noindex> a <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/asticode\/go-astideepspeech\">Go<\/a><\/noindex>). Hotov\u00fd model je dod\u00e1v\u00e1n pouze pro anglick\u00fd jazyk, ale pro jin\u00e9 jazyky lze podle <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/mozilla\/DeepSpeech\/wiki#how-can-i-train-using-my-own-data\">p\u0159ilo\u017een\u00e9ho<\/a><\/noindex> <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/mozilla\/DeepSpeech\/blob\/master\/TRAINING.rst#common-voice-training-data\">instrukci<\/a><\/noindex> n\u00e1vodu syst\u00e9m vytr\u00e9novat samostatn\u011b, za pou\u017eit\u00ed <noindex><a rel=\"nofollow\" href=\"https:\/\/voice.mozilla.org\/en\/datasets\">hlasov\u00fdch dat<\/a><\/noindex>, shrom\u00e1\u017ed\u011bn\u00fdch projektem Common Voice. <\/p>\n<p>DeepSpeech je v\u00fdrazn\u011b jednodu\u0161\u0161\u00ed ne\u017e tradi\u010dn\u00ed syst\u00e9my a z\u00e1rove\u0148 poskytuje vy\u0161\u0161\u00ed kvalitu rozpozn\u00e1v\u00e1n\u00ed p\u0159i p\u0159\u00edtomnosti ne\u017e\u00e1douc\u00edho \u0161umu. V procesu v\u00fdvoje nejsou pou\u017e\u00edv\u00e1ny tradi\u010dn\u00ed akustick\u00e9 modely a koncepty fon\u00e9m\u016f, m\u00edsto toho se uplat\u0148uje dob\u0159e optimalizovan\u00fd syst\u00e9m strojov\u00e9ho u\u010den\u00ed zalo\u017een\u00fd na neuronov\u00e9 s\u00edti, kter\u00fd umo\u017e\u0148uje obej\u00edt pot\u0159ebu vyv\u00edjet jednotliv\u00e9 komponenty pro modelov\u00e1n\u00ed r\u016fzn\u00fdch odchylek, jako jsou \u0161um, ozv\u011bna a rysy \u0159e\u010di. <\/p>\n<p>Nev\u00fdhodou tohoto p\u0159\u00edstupu je, \u017ee pro dosa\u017een\u00ed kvalitn\u00edho rozpozn\u00e1v\u00e1n\u00ed a tr\u00e9nov\u00e1n\u00ed neuronov\u00e9 s\u00edt\u011b motor DeepSpeech vy\u017eaduje velk\u00e9 mno\u017estv\u00ed r\u016fznorod\u00fdch dat, nadiktovan\u00fdch v re\u00e1ln\u00fdch podm\u00ednk\u00e1ch r\u016fzn\u00fdmi hlasy a za p\u0159\u00edtomnosti p\u0159irozen\u00e9ho \u0161umu.<br \/>\nSb\u00edrkou takov\u00fdch dat se zab\u00fdv\u00e1 projekt vytvo\u0159en\u00fd ve spole\u010dnosti Mozilla <noindex><a rel=\"nofollow\" href=\"https:\/\/voice.mozilla.org\/\">Common Voice<\/a><\/noindex>, kter\u00fd poskytuje ov\u011b\u0159en\u00fd datov\u00fd soubor s 780 hodinami v <noindex><a rel=\"nofollow\" href=\"https:\/\/voice.mozilla.org\/en\/datasets\">anglick\u00e9m jazyce<\/a><\/noindex>, 325 hodinami v n\u011bm\u010din\u011b, 173 hodinami ve francouz\u0161tin\u011b a 27 hodinami v ru\u0161tin\u011b. <\/p>\n<p>Kone\u010dn\u00fdm c\u00edlem projektu Common Voice je shrom\u00e1\u017edit 10 000 hodin nahr\u00e1vek r\u016fzn\u00fdch v\u00fdslovnost\u00ed b\u011b\u017en\u00fdch fr\u00e1z\u00ed lidsk\u00e9 \u0159e\u010di, co\u017e umo\u017en\u00ed dos\u00e1hnout p\u0159ijateln\u00fdch chybovosti p\u0159i rozpozn\u00e1v\u00e1n\u00ed. V sou\u010dasn\u00e9 podob\u011b \u00fa\u010dastn\u00edci projektu ji\u017e nadiktovali celkem 4,3 tis\u00edce hodin, z nich\u017e 3,5 tis\u00edce pro\u0161lo kontrolou. P\u0159i \u0161kolen\u00ed kone\u010dn\u00e9ho modelu anglick\u00e9ho jazyka pro DeepSpeech bylo vyu\u017eito 3816 hodin \u0159e\u010di, krom\u011b Common Voice pokr\u00fdvaj\u00edc\u00ed data z projekt\u016f LibriSpeech, Fisher a Switchboard, a zahrnuj\u00edc\u00ed p\u0159ibli\u017en\u011b 1700 hodin transkribovan\u00fdch nahr\u00e1vek rozhlasov\u00fdch po\u0159ad\u016f.<\/p>\n<p>P\u0159i pou\u017eit\u00ed nab\u00edzen\u00e9ho modelu anglick\u00e9ho jazyka pro sta\u017een\u00ed je \u00farove\u0148 chybovosti p\u0159i rozpozn\u00e1v\u00e1n\u00ed v DeepSpeech 7,5 % p\u0159i hodnocen\u00ed testovac\u00edm souborem. <noindex><a rel=\"nofollow\" href=\"http:\/\/www.openslr.org\/12\/\">LibriSpeech<\/a><\/noindex>. Pro porovn\u00e1n\u00ed, \u00farove\u0148 chybovosti p\u0159i rozpozn\u00e1v\u00e1n\u00ed \u010dlov\u011bkem <noindex><a rel=\"nofollow\" href=\"https:\/\/arxiv.org\/pdf\/1412.5567v1.pdf\">je hodnocena<\/a><\/noindex> je 5,83%.<\/p>\n<p>DeepSpeech se skl\u00e1d\u00e1 ze dvou subsyst\u00e9m\u016f \u2014 akustick\u00e9ho modelu a dekod\u00e9ru. Akustick\u00fd model vyu\u017e\u00edv\u00e1 metody hlubok\u00e9ho strojov\u00e9ho u\u010den\u00ed k v\u00fdpo\u010dtu pravd\u011bpodobnosti p\u0159\u00edtomnosti ur\u010dit\u00fdch symbol\u016f ve vstupn\u00edm zvuku. Dekod\u00e9r pou\u017e\u00edv\u00e1 algoritmus prohled\u00e1v\u00e1n\u00ed s v\u011btven\u00edm k p\u0159evodu dat o pravd\u011bpodobnosti symbol\u016f na textovou reprezentaci.<\/p>\n<p>Hlavn\u00ed <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/mozilla\/DeepSpeech\/releases\/v0.6.0\">novinky<\/a><\/noindex> DeepSpeech 0.6 (v\u011btev 0.6 nen\u00ed kompatibiln\u00ed se star\u0161\u00edmi verzemi a vy\u017eaduje aktualizaci k\u00f3du a model\u016f):<\/p>\n<ul>\n<li class=\"l\"> Byl navr\u017een nov\u00fd streamovac\u00ed dekod\u00e9r, kter\u00fd zaji\u0161\u0165uje vy\u0161\u0161\u00ed odezvu a nez\u00e1vis\u00ed na velikosti zpracov\u00e1van\u00fdch zvukov\u00fdch dat. V d\u016fsledku toho se v nov\u00e9 verzi DeepSpeech poda\u0159ilo sn\u00ed\u017eit latenci rozpozn\u00e1v\u00e1n\u00ed na 260 ms, co\u017e je o 73 % rychlej\u0161\u00ed ne\u017e d\u0159\u00edve, a umo\u017e\u0148uje pou\u017e\u00edvat DeepSpeech v \u0159e\u0161en\u00edch pro rozpozn\u00e1v\u00e1n\u00ed \u0159e\u010di v re\u00e1ln\u00e9m \u010dase.\n<li class=\"l\"> Byly provedeny zm\u011bny v API a byla provedena pr\u00e1ce na unifikaci n\u00e1zv\u016f funkc\u00ed. Byly p\u0159id\u00e1ny funkce pro z\u00edsk\u00e1v\u00e1n\u00ed dal\u0161\u00edch metadat o synchronizaci, co\u017e umo\u017e\u0148uje nejen z\u00edskat textovou reprezentaci, ale tak\u00e9 sledovat vazbu jednotliv\u00fdch symbol\u016f a v\u011bt na pozici v zvukov\u00e9m toku.\n<li class=\"l\"> Do n\u00e1strojov\u00e9ho bal\u00ed\u010dku pro \u0161kolen\u00ed modul\u016f byla p\u0159id\u00e1na podpora pro pou\u017eit\u00ed knihovny <noindex><a rel=\"nofollow\" href=\"https:\/\/developer.nvidia.com\/cudnn\">CuDNN<\/a><\/noindex> pro optimalizaci pr\u00e1ce s rekurentn\u00edmi neuronov\u00fdmi s\u00edt\u011bmi (RNN), co\u017e vedlo k v\u00fdrazn\u00e9mu (asi dvojn\u00e1sobn\u00e9mu) zv\u00fd\u0161en\u00ed v\u00fdkonnosti tr\u00e9nov\u00e1n\u00ed modelu, ale vy\u017eadovalo zm\u011bny v k\u00f3du, kter\u00e9 naru\u0161ily kompatibilitu s d\u0159\u00edve p\u0159ipraven\u00fdmi modely.\n<li class=\"l\"> Minim\u00e1ln\u00ed po\u017eadavky na verzi TensorFlow byly zv\u00fd\u0161eny z 1.13.1 na 1.14.0. Byla p\u0159id\u00e1na podpora lehk\u00e9 verze TensorFlow Lite, p\u0159i jej\u00edm\u017e pou\u017eit\u00ed se velikost bal\u00ed\u010dku DeepSpeech zmen\u0161ila z 98 MB na 3.7 MB. Pro pou\u017eit\u00ed na vestav\u011bn\u00fdch a mobiln\u00edch za\u0159\u00edzen\u00edch byl tak\u00e9 zmen\u0161en velikost zabalen\u00e9ho souboru s modelem z 188 MB na 47 MB (metoda kvantizace po dokon\u010den\u00ed tr\u00e9nov\u00e1n\u00ed modelu byla pou\u017eita pro kompresi).\n<li class=\"l\"> Jazykov\u00fd model byl p\u0159eveden na jin\u00fd form\u00e1t datov\u00fdch struktur, co\u017e umo\u017e\u0148uje prov\u00e1d\u011bt mapov\u00e1n\u00ed soubor\u016f do pam\u011bti p\u0159i na\u010d\u00edt\u00e1n\u00ed. Podpora star\u00e9ho form\u00e1tu byla ukon\u010dena.\n<li class=\"l\"> Byl zm\u011bn\u011bn re\u017eim na\u010d\u00edt\u00e1n\u00ed souboru s jazykov\u00fdm modelem, co\u017e vedlo ke sn\u00ed\u017een\u00ed spot\u0159eby pam\u011bti a zkr\u00e1cen\u00ed latence p\u0159i zpracov\u00e1n\u00ed prvn\u00edho po\u017eadavku po vytvo\u0159en\u00ed modelu. V procesu pr\u00e1ce DeepSpeech nyn\u00ed spot\u0159ebov\u00e1v\u00e1 22kr\u00e1t m\u00e9n\u011b pam\u011bti a spou\u0161t\u00ed se 500kr\u00e1t rychleji.\n<p><center><noindex><a rel=\"nofollow\" href=\"https:\/\/2r4s9p1yi1fa2jd7j43zph8r-wpengine.netdna-ssl.com\/files\/2019\/11\/startup_memory-1.png\"><img decoding=\"async\" alt=\"Spole\u010dnost Mozilla p\u0159edstavila engine pro rozpozn\u00e1v\u00e1n\u00ed \u0159e\u010di DeepSpeech 0.6.\" src=\"\/wp-content\/uploads\/2019\/12\/113ef11374da37383391cc886c488741.png\" style=\"display:block;margin: 0 auto;\" \/><\/a><\/noindex><\/center><\/p>\n<li class=\"l\"> Bylo provedeno filtrov\u00e1n\u00ed vz\u00e1cn\u00fdch slov v jazykov\u00e9m modelu. Celkov\u00fd po\u010det slov byl zredukov\u00e1n na 500 tis\u00edc nejpopul\u00e1rn\u011bj\u0161\u00edch slov, kter\u00e1 se vyskytuj\u00ed v textu pou\u017eit\u00e9m p\u0159i tr\u00e9nov\u00e1n\u00ed modelu. Proveden\u00e1 \u00faprava umo\u017enila sn\u00ed\u017eit velikost jazykov\u00e9ho modelu z 1800 MB na 900 MB, prakticky bez vlivu na ukazatele chybovosti rozpozn\u00e1v\u00e1n\u00ed.\n<li class=\"l\"> Byla p\u0159id\u00e1na podpora r\u016fzn\u00fdch <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/mozilla\/DeepSpeech\/commit\/58fdd55eea951dabd91a1158207f2228d745e657\">technik<\/a><\/noindex> vytv\u00e1\u0159en\u00ed dal\u0161\u00edch variac\u00ed (augmentation) zvukov\u00fdch dat, pou\u017e\u00edvan\u00fdch p\u0159i tr\u00e9nov\u00e1n\u00ed (nap\u0159\u00edklad p\u0159id\u00e1n\u00ed do sady variant, do kter\u00fdch byla zanesena zkreslen\u00ed nebo \u0161umy).\n<li class=\"l\"> Byla p\u0159id\u00e1na knihovna s vazbami pro integraci s aplikacemi na platform\u011b .NET.\n<li class=\"l\"> Byla p\u0159epracov\u00e1na dokumentace, kter\u00e1 je nyn\u00ed shrom\u00e1\u017ed\u011bna na samostatn\u00e9m webu <noindex><a rel=\"nofollow\" href=\"https:\/\/deepspeech.readthedocs.io\/en\/latest\">deepspeech.readthedocs.io<\/a><\/noindex>.\n<\/ul>\n<p><noindex><a rel=\"nofollow\" name=\"link\"><\/a><\/noindex><\/p>\n<p>Zdroj: <a \ncontent=\"nofollow\" rel=\"nofollow\" href=\"https:\/\/www.opennet.ru\/opennews\/art.shtml?num=51991\">opennet.ru<\/a><\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"<p>\u041f\u0440\u0435\u0434\u0441\u0442\u0430\u0432\u043b\u0435\u043d \u0432\u044b\u043f\u0443\u0441\u043a \u0440\u0430\u0437\u0432\u0438\u0432\u0430\u0435\u043c\u043e\u0433\u043e \u043a\u043e\u043c\u043f\u0430\u043d\u0438\u0435\u0439 Mozilla \u0434\u0432\u0438\u0436\u043a\u0430 \u0440\u0430\u0441\u043f\u043e\u0437\u043d\u0430\u0432\u0430\u043d\u0438\u044f \u0440\u0435\u0447\u0438 DeepSpeech 0.6, \u043a\u043e\u0442\u043e\u0440\u044b\u0439 \u0440\u0435\u0430\u043b\u0438\u0437\u0443\u0435\u0442 \u043e\u0434\u043d\u043e\u0438\u043c\u0451\u043d\u043d\u0443\u044e \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440\u0443 \u0440\u0430\u0441\u043f\u043e\u0437\u043d\u0430\u0432\u0430\u043d\u0438\u044f \u0440\u0435\u0447\u0438, \u043f\u0440\u0435\u0434\u043b\u043e\u0436\u0435\u043d\u043d\u0443\u044e \u0438\u0441\u0441\u043b\u0435\u0434\u043e\u0432\u0430\u0442\u0435\u043b\u044f\u043c\u0438 \u0438\u0437 \u043a\u043e\u043c\u043f\u0430\u043d\u0438\u0438 Baidu. \u0420\u0435\u0430\u043b\u0438\u0437\u0430\u0446\u0438\u044f \u043d\u0430\u043f\u0438\u0441\u0430\u043d\u0430 \u043d\u0430 \u044f\u0437\u044b\u043a\u0435 Python \u0441 \u0438\u0441\u043f\u043e\u043b\u044c\u0437\u043e\u0432\u0430\u043d\u0438\u0435\u043c \u043f\u043b\u0430\u0442\u0444\u043e\u0440\u043c\u044b \u043c\u0430\u0448\u0438\u043d\u043d\u043e\u0433\u043e \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u044f TensorFlow \u0438 \u0440\u0430\u0441\u043f\u0440\u043e\u0441\u0442\u0440\u0430\u043d\u044f\u0435\u0442\u0441\u044f \u043f\u043e\u0434 \u0441\u0432\u043e\u0431\u043e\u0434\u043d\u043e\u0439 \u043b\u0438\u0446\u0435\u043d\u0437\u0438\u0435\u0439 MPL 2.0. \u041f\u043e\u0434\u0434\u0435\u0440\u0436\u0438\u0432\u0430\u0435\u0442\u0441\u044f \u0440\u0430\u0431\u043e\u0442\u0430 \u0432 Linux, Android, macOS \u0438 Windows. \u041f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0438\u0442\u0435\u043b\u044c\u043d\u043e\u0441\u0442\u0438 \u0434\u043e\u0441\u0442\u0430\u0442\u043e\u0447\u043d\u043e \u0434\u043b\u044f \u0438\u0441\u043f\u043e\u043b\u044c\u0437\u043e\u0432\u0430\u043d\u0438\u044f \u0434\u0432\u0438\u0436\u043a\u0430 \u043d\u0430 \u043f\u043b\u0430\u0442\u0430\u0445 LePotato, [&hellip;]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":53723,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[702],"tags":[],"class_list":["post-53722","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news"],"aioseo_notices":[],"aioseo_head":"\n\t\t<!-- All in One SEO 5.0.3 - aioseo.com -->\n\t<meta name=\"description\" content=\"\u041f\u0440\u0435\u0434\u0441\u0442\u0430\u0432\u043b\u0435\u043d \u0432\u044b\u043f\u0443\u0441\u043a \u0440\u0430\u0437\u0432\u0438\u0432\u0430\u0435\u043c\u043e\u0433\u043e \u043a\u043e\u043c\u043f\u0430\u043d\u0438\u0435\u0439 Mozilla \u0434\u0432\u0438\u0436\u043a\u0430 \u0440\u0430\u0441\u043f\u043e\u0437\u043d\u0430\u0432\u0430\u043d\u0438\u044f \u0440\u0435\u0447\u0438\" \/>\n\t<meta name=\"robots\" content=\"max-image-preview:large\" \/>\n\t<meta name=\"author\" content=\"Yuri Gagarin\"\/>\n\t<link rel=\"canonical\" href=\"https:\/\/prohoster.info\/cs\/blog\/news\/kompaniya-mozilla-predstavila-dvizhok-raspoznavaniya-rechi-deepspeech-0-6\" \/>\n\t\t<meta name=\"generator\" content=\"All in One SEO (AIOSEO) 5.0.3\" \/>\n\t\t<meta property=\"og:locale\" content=\"cs_CZ\" \/>\n\t\t<meta property=\"og:site_name\" content=\"ProHoster | \u041a\u0443\u043f\u0438\u0442\u044c \u043d\u0430\u0434\u0435\u0436\u043d\u044b\u0439 \u0445\u043e\u0441\u0442\u0438\u043d\u0433 \u0434\u043b\u044f \u0441\u0430\u0439\u0442\u043e\u0432 \u0441 \u0437\u0430\u0449\u0438\u0442\u043e\u0439 \u043e\u0442 DDoS, VPS VDS \u0441\u0435\u0440\u0432\u0435\u0440\u044b\" \/>\n\t\t<meta property=\"og:type\" content=\"article\" \/>\n\t\t<meta property=\"og:title\" content=\"\ud83e\udd47\u041a\u043e\u043c\u043f\u0430\u043d\u0438\u044f Mozilla \u043f\u0440\u0435\u0434\u0441\u0442\u0430\u0432\u0438\u043b\u0430 \u0434\u0432\u0438\u0436\u043e\u043a \u0440\u0430\u0441\u043f\u043e\u0437\u043d\u0430\u0432\u0430\u043d\u0438\u044f \u0440\u0435\u0447\u0438 DeepSpeech 0.6 | ProHoster\" \/>\n\t\t<meta property=\"og:description\" content=\"\u041f\u0440\u0435\u0434\u0441\u0442\u0430\u0432\u043b\u0435\u043d \u0432\u044b\u043f\u0443\u0441\u043a \u0440\u0430\u0437\u0432\u0438\u0432\u0430\u0435\u043c\u043e\u0433\u043e \u043a\u043e\u043c\u043f\u0430\u043d\u0438\u0435\u0439 Mozilla \u0434\u0432\u0438\u0436\u043a\u0430 \u0440\u0430\u0441\u043f\u043e\u0437\u043d\u0430\u0432\u0430\u043d\u0438\u044f \u0440\u0435\u0447\u0438\" \/>\n\t\t<meta property=\"og:url\" content=\"https:\/\/prohoster.info\/cs\/blog\/news\/kompaniya-mozilla-predstavila-dvizhok-raspoznavaniya-rechi-deepspeech-0-6\" \/>\n\t\t<meta property=\"og:image\" content=\"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg\" \/>\n\t\t<meta property=\"og:image:secure_url\" content=\"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg\" \/>\n\t\t<meta property=\"og:image:width\" content=\"350\" \/>\n\t\t<meta property=\"og:image:height\" content=\"350\" \/>\n\t\t<meta property=\"article:published_time\" content=\"2019-12-07T21:00:00+00:00\" \/>\n\t\t<meta property=\"article:modified_time\" content=\"2020-02-18T11:01:39+00:00\" \/>\n\t\t<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/prohoster\" \/>\n\t\t<meta property=\"article:author\" content=\"https:\/\/www.facebook.com\/prohoster\" \/>\n\t\t<!-- All in One SEO -->\n\n","aioseo_head_json":{"title":"\ud83e\udd47Spole\u010dnost Mozilla p\u0159edstavila rozpozn\u00e1vac\u00ed engine DeepSpeech 0.6 | ProHoster","description":"Bylo p\u0159edstaveno vyd\u00e1n\u00ed vyv\u00edjen\u00e9ho rozpozn\u00e1vac\u00edho enginu spole\u010dnosti Mozilla","canonical_url":"https:\/\/prohoster.info\/cs\/blog\/news\/kompaniya-mozilla-predstavila-dvizhok-raspoznavaniya-rechi-deepspeech-0-6","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"cs_CZ","og:site_name":"ProHoster | \u041a\u0443\u043f\u0438\u0442\u044c \u043d\u0430\u0434\u0435\u0436\u043d\u044b\u0439 \u0445\u043e\u0441\u0442\u0438\u043d\u0433 \u0434\u043b\u044f \u0441\u0430\u0439\u0442\u043e\u0432 \u0441 \u0437\u0430\u0449\u0438\u0442\u043e\u0439 \u043e\u0442 DDoS, VPS VDS \u0441\u0435\u0440\u0432\u0435\u0440\u044b","og:type":"article","og:title":"\ud83e\udd47\u041a\u043e\u043c\u043f\u0430\u043d\u0438\u044f Mozilla \u043f\u0440\u0435\u0434\u0441\u0442\u0430\u0432\u0438\u043b\u0430 \u0434\u0432\u0438\u0436\u043e\u043a \u0440\u0430\u0441\u043f\u043e\u0437\u043d\u0430\u0432\u0430\u043d\u0438\u044f \u0440\u0435\u0447\u0438 DeepSpeech 0.6 | ProHoster","og:description":"\u041f\u0440\u0435\u0434\u0441\u0442\u0430\u0432\u043b\u0435\u043d \u0432\u044b\u043f\u0443\u0441\u043a \u0440\u0430\u0437\u0432\u0438\u0432\u0430\u0435\u043c\u043e\u0433\u043e \u043a\u043e\u043c\u043f\u0430\u043d\u0438\u0435\u0439 Mozilla \u0434\u0432\u0438\u0436\u043a\u0430 \u0440\u0430\u0441\u043f\u043e\u0437\u043d\u0430\u0432\u0430\u043d\u0438\u044f \u0440\u0435\u0447\u0438","og:url":"https:\/\/prohoster.info\/cs\/blog\/news\/kompaniya-mozilla-predstavila-dvizhok-raspoznavaniya-rechi-deepspeech-0-6","og:image":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:secure_url":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:width":350,"og:image:height":350,"article:published_time":"2019-12-07T21:00:00+00:00","article:modified_time":"2020-02-18T11:01:39+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"53722","title":null,"description":null,"keywords":null,"keyphrases":null,"primary_term":null,"canonical_url":null,"og_title":null,"og_description":null,"og_object_type":"default","og_image_type":"default","og_image_url":null,"og_image_width":null,"og_image_height":null,"og_image_custom_url":null,"og_image_custom_fields":null,"og_video":null,"og_custom_url":null,"og_article_section":null,"og_article_tags":null,"twitter_use_og":false,"twitter_card":"default","twitter_image_type":"default","twitter_image_url":null,"twitter_image_custom_url":null,"twitter_image_custom_fields":null,"twitter_title":null,"twitter_description":null,"schema":{"blockGraphs":[],"customGraphs":[],"default":{"data":{"Article":[],"Course":[],"Dataset":[],"FAQPage":[],"Movie":[],"Person":[],"Product":[],"ProductReview":[],"Car":[],"Recipe":[],"Service":[],"SoftwareApplication":[],"WebPage":[]},"graphName":"","isEnabled":true},"graphs":[]},"schema_type":null,"schema_type_options":null,"pillar_content":false,"robots_default":true,"robots_noindex":false,"robots_noarchive":false,"robots_nosnippet":false,"robots_nofollow":false,"robots_noimageindex":false,"robots_noodp":false,"robots_notranslate":false,"robots_max_snippet":null,"robots_max_videopreview":null,"robots_max_imagepreview":"large","priority":null,"frequency":null,"local_seo":null,"seo_analyzer_scan_date":"2026-01-24 08:28:20","breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-02-28 20:19:21","updated":"2026-01-24 08:28:20","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/cs\/wp-json\/wp\/v2\/posts\/53722","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/cs\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/cs\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/cs\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/cs\/wp-json\/wp\/v2\/comments?post=53722"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/cs\/wp-json\/wp\/v2\/posts\/53722\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/prohoster.info\/cs\/wp-json\/wp\/v2\/media\/53723"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/cs\/wp-json\/wp\/v2\/media?parent=53722"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/cs\/wp-json\/wp\/v2\/categories?post=53722"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/cs\/wp-json\/wp\/v2\/tags?post=53722"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}