SpeakScribe has three Korean voices: SunHi (female), and InJoon and Hyunsu (male). SunHi is labelled as standard Seoul Korean. Hyunsu is multilingual, so one narrator can read Korean and English in the same script.
How Korean is written, and what to type so the voice reads it right
Hangul is an alphabet whose letters are grouped into syllable blocks, and word spacing (띄어쓰기) is part of correct spelling. A missing or extra space can change the meaning of a sentence as well as its rhythm. Type the script with the spacing a Korean editor would use, and run a spell checker that includes spacing rules before you generate.
Korean sentence endings show the speech level. Choose one level for the whole script.
Spacing changes meaning: 아버지가 방에 들어가신다 (father enters the room) and 아버지 가방에 들어가신다 (father gets into the bag) differ by one space.
Write the formal -습니다 ending for news, announcements and corporate videos, and the polite -요 ending for vlogs, tutorials and storytelling.
Do not mix -습니다 and -요 in one script. The voice reads each ending as written, so mixed levels sound inconsistent.
Write loanwords the way Koreans spell them, such as 유튜브 and 스마트폰, and keep Latin letters only for brand names you want read in English.
Numbers, dates and loanwords in Korean text
Korean uses two number systems. Native Korean numbers are used for hours and ages (세 시, 스무 살), and Sino-Korean numbers for minutes and money (삼십 분, 오천 원). If a number is read in the wrong system, write it in Hangul.
Dates are written with 년, 월 and 일, as in 2026년 10월 22일, which keeps each part of the date clearly marked.
Write hours with native numbers and minutes with Sino-Korean numbers: 세 시 and 십오 분 (15 minutes).
Write currency with a Sino-Korean number and 원: 오천 원.
Modern Korean is written in Hangul; Hanja (Chinese characters) are rarely used in everyday writing.
Accents and regional varieties
Korean has regional dialects, such as the Busan and Jeju varieties. None of the three voices is labelled with a regional accent in the library, and SunHi is described as standard Seoul Korean, so write standard Korean for these voices.
Preview any dialect line before you publish, since none of the three voices is labelled as a dialect speaker.
Choosing between the Korean voices
SunHi (female, 205.3 Hz median pitch, 11.0 semitones of intonation range) has the widest melodic range of the three. InJoon (male, 149.3 Hz, 8.8 semitones) has the brightest timbre, with a spectral centroid of 2,389 Hz. Hyunsu (male, 136.7 Hz, 8.1 semitones) is the lowest in pitch and has the narrowest range.
Choose Hyunsu when the same narrator reads Korean and English lines, and see the multilingual voices page for other voices that read several languages. To add a pause, use a comma or a full stop at the point where the pause belongs.
Common mistakes when turning Korean text into speech
Korean scripts usually break because of spacing, keyboard input or internet shorthand. These fixes apply before any voice is chosen.
Typing jamo separately, such as ㅎㅏㄴ, instead of composed syllables such as 한. Spell-checkers and spacing rules work on composed syllables, so retype the word.
Confusing 안 (not, placed before a verb) with -지 않다 (the negative ending): 안 가요 and 가지 않아요 are both correct, but they are not interchangeable in every sentence.
Writing 할 수 있다 as 할수있다. Dependent nouns such as 수 take a space before them.
Using internet shorthand such as ㅋㅋ or ㅠㅠ in a script. Write the words out, since the voice reads the letters you type.
Writing Hanja for everyday words that Koreans write in Hangul. Use Hanja only where the term needs it.
Korean voices compared — measured
Each voice read the same Korean text at normal speed. Lower pitch sounds deeper; a wider intonation range sounds livelier, a narrower one steadier.
Measured by SpeakScribe in October 2026 from each voice's audio. Pitch is the median fundamental frequency; intonation range is the spread between the 10th and 90th percentile of pitch. Speaking rate is characters per second, spaces included, on the ~15-second sample passage that every Korean voice reads. st = semitones. Methodology · How SpeakScribe voices work