Означе изабрани модел разуме — кликните да га испустите у текст где се то дешава:
Овај модел чита обичан текст, па се игноришу ознаке у пољима. За емоције на основу ознака, пребаците се на изражавање модела попут Orpheus или Bark.
Upotreba prirodnog jezika za davanje instrukcija. Zahvaljujemo se korisniku Qwen3-TTS danas; uskoro će biti dostupni i drugi modeli sa naprednim mogućnostima.
Овај модел не подржава упутства у стилу — пребаци на модел који подржава (нпр. Qwen3-TTS).
Дефинишите посебне изговоре (слов = изговор):
-12
+12
Контролише експресивност (0 = неутрална, 1 = веома изразита)
Тежина навођења без класификатора (више = више пратеће)
Опиши стил гласа на природном језику (Парлер користи ово уместо предефинисаних гласова)
Формат дија‐ дијалога: Користите [S1] и [S2] ознаке за различите звучнике. Прим› јер: [S1] Здраво! [S2] Здраво, како сте?
Greek text-to-speech works in the Greek alphabet — a 24-letter script that is the ancestor of both Latin and Cyrillic and shares no characters with them, so the model reads native Ελληνικά rather than transliteration. Modern Greek pronunciation is largely phonemic but has rich digraph and diphthong rules: pairs like ου read as "u", αι as "e", ει/οι/υ as "i" (the result of historical iotacism, where several spellings collapsed onto the same vowel sound), and μπ, ντ, γκ stand in for the b, d, and g sounds. Stress is marked explicitly by the tonos accent (ά, έ, ή, ί, ό, ύ, ώ) and is phonemic — "πότε" (when) versus "ποτέ" (never) differ only in where the stress falls — so a synthesizer must honor the written accent to say the right word. Distinctive sounds like the two θ/δ "th" fricatives, the velar/palatal split of γ and χ, and final-n sandhi all require a model built specifically for Greek phonology.
Узорак — Ελληνικά
“Η τεχνολογία μετατροπής κειμένου σε ομιλία διαβάζει τις ελληνικές προτάσεις με φυσική φωνή, ώστε να ακούγονται εύκολα.”
Родно име
Ελληνικά
Звучници
About 13.5 million speakers; the official language of Greece and Cyprus and an official language of the EU.
Језичка породица
Indo-European (Hellenic — its own independent branch)
Скрипта
Greek alphabet (24 letters), the source of Latin and Cyrillic
Причао сам.
Greece, Cyprus, and Greek diaspora communities in the US, Australia, and across Europe
слободниFemaleСамо за особне и некомерцијалне потребе
Шта људи користе Greek текст за говор
Greek audiobook and literary text narration
Accessibility screen-reading for Greek and Cypriot visually impaired users
IVR and customer-service phone prompts for Greek businesses
E-learning, museum-guide, and tourism voiceover in Greek
Localized app and e-commerce voice prompts for the Greek and Cypriot markets
Greek Текст у говор — честа питања
Yes. Paste your text in native Greek script (Ελληνικά) — the 24-letter Greek alphabet — and the model reads it directly. You do not need to transliterate into Latin letters ("Greeklish"), which would degrade pronunciation and accent placement.
It does, and this matters: Greek stress is phonemic, so "πότε" (when) and "ποτέ" (never), or "νόμος" (law) and "νομός" (province), differ only by accent position. Keep the tonos marks in your text so the voice stresses the correct syllable and says the intended word.
The model applies Modern Greek reading rules: ου → "u", αι → "e", ει/οι → "i", and the clusters μπ, ντ, γκ produce b, d, and g sounds (e.g. "μπύρα" → "beera"). This reflects iotacism and modern phonology rather than ancient or letter-by-letter readings.
Yes. Greek has two "th" letters — θ (as in "thin") and δ (as in "this") — plus γ and χ, which shift between velar and palatal sounds depending on the following vowel. The voice is trained on native Greek phonology so these fricatives are rendered distinctly rather than approximated.