Narrated channels without a studio
Publishing daily without recording in a treated room.
Narration, character work and YouTube in Japanese. The kanji and pitch problems are explained here rather than glossed over.
Free plan · No card · Japanese voices · MP3 and MP4 export
Engines behind the Japanese voices
Several engines supply ja-JP voices here, of the engines Fliki runs. Switch line by line and keep whichever reads your names correctly.
3 TTS engines
All Japanese AI voices
Every voice tagged to this dialect, across each language and TTS engine that covers it. Many have an instant sample you can tap to listen to, no account needed.
Aoi (Child)♀
Standard
Daichi♂
Standard
Ishibashi♀
Multilingual
Kazuha♀
Standard
Keita♂
Standard
Masaru♂
Ultra-Realistic
Mayu♀
Standard
Mizuki♀
Standard
Nanami♀
Ultra-Realistic
Naoki♂
Standard
Shiori♀
Standard
Takumi♂
Standard
Tomoko♀
Standard
Two things that break Japanese narration
Japanese has one standard register and two genuine ways for a read to go wrong.
A single character takes a native reading in one word and a Chinese-derived one in another, and names are worse: the same two characters can be Nakata or Nakada with nothing in the text to decide it. Engines pick by frequency and are wrong often enough on names, places and technical terms to matter. Writing the reading in kana beside the first occurrence solves it.
Hashi is a bridge or chopsticks depending on where the pitch falls, and nothing in the spelling marks it. A wrong pitch is not a wrong sound; it is a different word, and a Japanese listener hears it instantly even when the sentence still parses. Context usually saves the engine. Where it does not, the fix is to rephrase rather than to change voices.
How it works
Write it, test the names, set the register, export.
Japanese sentence marks drive the phrasing, so a script stripped of them runs together. Keep the full stops and commas as typed rather than converting them to Latin ones.
Personal names, place names and company names are where kanji readings go wrong. Put them through the box at the top of this page first, and write the kana reading in the script for any that come out incorrectly.
Desu-masu for corporate and instructional material, plain form for casual video. The register is a property of the script, not of the voice, and a polite voice reading plain-form text does not make it polite.
Download MP3 for podcasts, e-learning and announcements, or build visuals and Japanese subtitles here and export a 1080p MP4.
What Japanese AI voices are used for
The Japanese content being produced on a schedule.
Publishing daily without recording in a treated room.
Scratch dialogue and prototypes before a cast is booked.
Material that gets revised more often than it gets watched.
A single edit revoiced rather than reshot for the market.
Short recordings updated on a fixed schedule.
Japanese voice generator FAQ
Kanji readings, pitch accent, anime voices, cost and rights.
AI voices by language
Tools
Japanese voices, kanji readings you can control, MP3 and MP4 export. No card.
Generate a Japanese voice freeFree forever plan · No credit card required · Cancel anytime