{"id":102522,"date":"2021-12-08T15:37:00","date_gmt":"2021-12-08T13:37:00","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/vypusk-sintezatora-rechi-rhvoice-1-6-0"},"modified":"2021-12-08T15:37:00","modified_gmt":"2021-12-08T13:37:00","slug":"vypusk-sintezatora-rechi-rhvoice-1-6-0","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sq\/blog\/news\/vypusk-sintezatora-rechi-rhvoice-1-6-0","title":{"rendered":"L\u00ebshimi i sintezatorit t\u00eb z\u00ebrit RHVoice 1.6.0","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
L\u00ebshimi i sistemit t\u00eb hapur t\u00eb sintez\u00ebs s\u00eb z\u00ebrit RHVoice 1.6.0, fillimisht i zhvilluar p\u00ebr t\u00eb siguruar mb\u00ebshtetje cil\u00ebsore p\u00ebr gjuh\u00ebn ruse, por m\u00eb pas i adaptuar edhe p\u00ebr gjuh\u00eb t\u00eb tjera, p\u00ebrfshir\u00eb anglishten, portugezin, ukrainishten, kirgizishten, tatarishten dhe gjeorgjishten. Kodi \u00ebsht\u00eb shkruar n\u00eb C++ dhe distribuonhet n\u00ebn licenc\u00ebn LGPL 2.1. Mb\u00ebshtetet puna n\u00eb GNU\/Linux, Windows dhe Android. Programi \u00ebsht\u00eb i pajtuesh\u00ebm me standartet e zakonshme TTS (text-to-speech) p\u00ebr konvertimin e tekstit n\u00eb z\u00eb: SAPI5 (Windows), Speech Dispatcher (GNU\/Linux) dhe Android Text-To-Speech API, por mund t\u00eb p\u00ebrdoret gjithashtu n\u00eb lexuesin e ekranit NVDA. Krijuesja dhe zhvilluesja kryesore e RHVoice \u00ebsht\u00eb Olga Yakovleva, e cila vazhdon t\u00eb zhvilloj\u00eb projektin pavar\u00ebsisht nga verb\u00ebria e plot\u00eb. <\/p>\n
N\u00eb versionin e ri jan\u00eb shtuar 5 variante t\u00eb reja z\u00ebri p\u00ebr t\u00eb folurimin rus. \u00cbsht\u00eb realizuar mb\u00ebshtetje p\u00ebr gjuh\u00ebn shqipe. Fjalori p\u00ebr gjuh\u00ebn ukrainase \u00ebsht\u00eb p\u00ebrdit\u00ebsuar. Mb\u00ebshtetje p\u00ebr shqiptime t\u00eb simboleve emoji \u00ebsht\u00eb zgjeruar. Jan\u00eb b\u00ebr\u00eb p\u00ebrmir\u00ebsime p\u00ebr t\u00eb eliminuar gabimet n\u00eb aplikacionin p\u00ebr platform\u00ebn Android, importimi i fjalor\u00ebve individual \u00ebsht\u00eb thjeshtuar, dhe \u00ebsht\u00eb shtuar mb\u00ebshtetje p\u00ebr platform\u00ebn Android 11. N\u00eb b\u00ebrtham\u00ebn e motorit jan\u00eb shtuar cil\u00ebsime dhe funksionalitete t\u00eb reja, duke p\u00ebrfshir\u00eb g2p.case, word_break dhe mb\u00ebshtetje p\u00ebr filtre ekuilizimi. <\/p>\n
Kujtojm\u00eb se n\u00eb RHVoice p\u00ebrdoren arritjet e projektit HTS (HMM\/DNN-based Speech Synthesis System) dhe metoda parametrike e sintez\u00ebs me modelet statistikore (Statistical Parametric Synthesis mbi baz\u00ebn e HMM - Hidden Markov Model). Avantazhi i modelit statistik \u00ebsht\u00eb se ka shpenzime t\u00eb ul\u00ebta operacionale dhe nuk k\u00ebrkon shum\u00eb energji nga CPU. T\u00eb gjitha operacionet kryhen lokal n\u00eb sistemin e p\u00ebrdoruesit. Mb\u00ebshteten tre nivele t\u00eb cil\u00ebsis\u00eb s\u00eb z\u00ebrit (sa m\u00eb i ul\u00ebt t\u00eb jet\u00eb niveli i cil\u00ebsis\u00eb - aq m\u00eb e lart\u00eb \u00ebsht\u00eb performanca dhe m\u00eb e shpejt\u00eb \u00ebsht\u00eb koha e reagimit). <\/p>\n
Nj\u00eb nga disavantazhet e modelit statistik \u00ebsht\u00eb cil\u00ebsia e ul\u00ebt e shqiptimit, e cila nuk arrin nivelin e sintetizator\u00ebve q\u00eb gjenerojn\u00eb z\u00eb me kombinim t\u00eb fragmenteve t\u00eb z\u00ebrit natyror. Megjithat\u00eb, rezultati \u00ebsht\u00eb mjaft i qart\u00eb dhe p\u00ebrngjan me transmetimin e nj\u00eb regjistrimi nga nj\u00eb megafon. Si krahasim, projekti Silero, i cili ofron nj\u00eb motor t\u00eb hapur p\u00ebr sintetizimin e z\u00ebrit duke p\u00ebrdorur teknologji t\u00eb m\u00ebsimit t\u00eb makinave dhe nj\u00eb grup modelesh p\u00ebr gjuh\u00ebn ruse, \u00ebsht\u00eb m\u00eb i mir\u00eb se RHVoice n\u00eb cil\u00ebsi. <\/p>\n
P\u00ebr gjuh\u00ebn ruse jan\u00eb t\u00eb disponuesh\u00ebm 13 variante z\u00ebresh, p\u00ebr anglishten — 5. Z\u00ebrat formohen mbi baz\u00ebn e regjistrimeve t\u00eb fjal\u00ebs s\u00eb natyrshme. N\u00eb cil\u00ebsime, mund t\u00eb ndryshoni shpejt\u00ebsin\u00eb, lart\u00ebsin\u00eb dhe volumin. P\u00ebr ndryshimin e ritmit mund t\u00eb p\u00ebrdoret biblioteka Sonic. \u00cbsht\u00eb e mundur t\u00eb identifikohet dhe t\u00eb kaloj\u00eb automatikisht n\u00eb gjuh\u00eb n\u00eb baz\u00eb t\u00eb analiz\u00ebs s\u00eb tekstit t\u00eb hyrjes (p\u00ebr shembull, p\u00ebr fjal\u00eb dhe citime n\u00eb nj\u00eb gjuh\u00eb tjet\u00ebr mund t\u00eb p\u00ebrdoret modeli i sintez\u00ebs natyrore p\u00ebr k\u00ebt\u00eb gjuh\u00eb). Mb\u00ebshteten profilet e z\u00ebrit, q\u00eb p\u00ebrcaktojn\u00eb kombinimet e z\u00ebrave p\u00ebr gjuh\u00eb t\u00eb ndryshme. \n Burimi: opennet.ru<\/a> <\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"