Dutch Матндан сўзга
Ўгир Dutch AI овозлари билан матнни табиий сўзлашувга айлантиринг. 2 овозлар. Бепул, рўйхатдан ўтиш керак эмас — MP3 ёки WAV форматида юклаб олиш мумкин.
Матнни аниқ назорат учун SSML теглар билан ўраб қўйиш:
<speak><prosody rate="slow">Slow speech</prosody></speak>
Танланган модел тушунадиган тегилар — уларни матнга тушириш учун босинг:
Бу модел оддий матн ўқийди, шунинг учун тегишли теги эътиборга олинмайди. Эмоционал теги асосида, Orpheus ёки Bark каби ифодали моделга ўтинг.
Ўз нутқини белгилаш (сўз = нутқ):
Маълумот Dutch Матндан сўзга
Dutch text-to-speech must master a set of guttural and diphthong sounds that have no close English equivalent: the hard "g" and "ch" fricatives, and the rolling diphthongs in "ui", "ij", and "eui". The same orthography is shared by two main standards — Netherlands Dutch and Belgian Flemish — which differ noticeably in rhythm and the realization of the "g", so a voice trained on one sounds foreign to the other. Dutch also forms long compound nouns and uses many spelling-rule vowel doublings (aa, ee, oo) that signal length, all of which the engine must parse for correct stress. The frequent assimilation and final-devoicing at word boundaries add a further layer the synthesizer has to model to avoid a clipped, list-like delivery.
Намуна — Nederlands
“Met tekst-naar-spraak laat je de computer Nederlandse teksten voorlezen met een natuurlijke, heldere stem.”
- Туғилган исми
- Nederlands
- Овозли қулфлар
- ~25 million native speakers
- Тил оиласи
- West Germanic (Indo-European)
- Скрипт
- Latin
- Сўзланган
- Netherlands, Belgium (Flanders), Suriname, Caribbean
