{"id":100106,"date":"2021-05-06T10:22:50","date_gmt":"2021-05-06T08:22:50","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/vypusk-sintezatora-rechi-rhvoice-1-2-4-razvivaemogo-dlya-russkogo-yazyka"},"modified":"2021-05-06T10:22:50","modified_gmt":"2021-05-06T08:22:50","slug":"vypusk-sintezatora-rechi-rhvoice-1-2-4-razvivaemogo-dlya-russkogo-yazyka","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sk\/blog\/news\/vypusk-sintezatora-rechi-rhvoice-1-2-4-razvivaemogo-dlya-russkogo-yazyka","title":{"rendered":"Vydanie syntetiz\u00e1tora re\u010di RHVoice 1.2.4, vyv\u00edjan\u00e9ho pre rusk\u00fd jazyk","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
Bolestn\u00fd syst\u00e9m synt\u00e9zy re\u010di RHVoice 1.2.4 bol zverejnen\u00fd, p\u00f4vodne vyvinut\u00fd na zabezpe\u010denie kvalitnej podpory pre ru\u0161tinu, ale n\u00e1sledne upraven\u00fd aj pre \u010fal\u0161ie jazyky, vr\u00e1tane angli\u010dtiny, portugal\u010diny, ukrajin\u010diny, kirgiz\u010diny, tat\u00e1r\u010diny a gruz\u00edn\u010diny. K\u00f3d je nap\u00edsan\u00fd v C++ a \u0161\u00edri sa pod licenciou LGPL 2.1. Podporuje pr\u00e1cu v GNU\/Linux, Windows a Android. Program je kompatibiln\u00fd so \u0161tandardn\u00fdmi TTS rozhraniami (text-to-speech) na prevod textu na re\u010d: SAPI5 (Windows), Speech Dispatcher (GNU\/Linux) a Android Text-To-Speech API, ale m\u00f4\u017ee sa pou\u017ei\u0165 aj v ekranovom \u010d\u00edta\u010di NVDA. <\/p>\n
V programe sa pou\u017e\u00edva parametrick\u00e1 met\u00f3da synt\u00e9zy so \u0161tatistick\u00fdmi modelmi (Statistical Parametric Synthesis na b\u00e1ze HMM \u2014 Hidden Markov Model). V\u00fdhodou \u0161tatistick\u00e9ho modelu s\u00fa n\u00edzke prev\u00e1dzkov\u00e9 n\u00e1klady a nen\u00e1ro\u010dnos\u0165 na v\u00fdkon CPU. V\u0161etky oper\u00e1cie sa vykon\u00e1vaj\u00fa lok\u00e1lne v syst\u00e9me pou\u017e\u00edvate\u013ea. Podporuj\u00fa sa tri \u00farovne kvality re\u010di (\u010d\u00edm ni\u017e\u0161ia je kvalita \u2014 t\u00fdm vy\u0161\u0161\u00ed je v\u00fdkon a ni\u017e\u0161ia latencia). <\/p>\n
Podporuje sa nastavenie a zmena hlasov. Pre ru\u0161tinu je k dispoz\u00edcii 9 variantov hlasov, pre angli\u010dtinu \u2014 5. Hlasov\u00e9 prejavy sa vytv\u00e1raj\u00fa na z\u00e1klade nahr\u00e1vok prirodzenej re\u010di. Vzh\u013eadom na pou\u017eitie \u0161tatistick\u00e9ho modelu kvalita v\u00fdslovnosti nedosahuje \u00farove\u0148 syntetiz\u00e1torov, ktor\u00e9 generuj\u00fa re\u010d na z\u00e1klade kombin\u00e1cie fragmentov prirodzenej re\u010di, ale v\u00fdsledok je st\u00e1le dostato\u010dne zrozumite\u013en\u00fd a pripom\u00edna vysielanie nahr\u00e1vky z reproduktora. <\/p>\n
V nastaveniach mo\u017eno meni\u0165 r\u00fdchlos\u0165, v\u00fd\u0161ku a hlasitos\u0165. Na \u00fapravu tempa m\u00f4\u017ee by\u0165 pou\u017eit\u00e1 kni\u017enica Sonic. Je mo\u017en\u00e9 automatick\u00e9 ur\u010dovanie a prepolenie jazyka na z\u00e1klade anal\u00fdzy vstupn\u00e9ho textu (napr\u00edklad pre slov\u00e1 a cit\u00e1ty v inom jazyku m\u00f4\u017ee by\u0165 pou\u017eit\u00fd dom\u00e1ci model synt\u00e9zy pre dan\u00fd jazyk). Podporuj\u00fa sa hlasov\u00e9 profily, ktor\u00e9 ur\u010duj\u00fa kombin\u00e1cie hlasov pre r\u00f4zne jazyky. \n Zdroj: opennet.ru<\/a> <\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"