AnyVoice turns Japanese text to speech in seconds — free, no account, nothing stored. Type kanji, kana or mixed text, pick a male or female voice, and download the MP3. It converts text into audio, not the reverse.
Japanese has one standard accent, so the real choice is register: a voice that suits a podcast rarely suits a support line. Pick the job first, then the voice.
The free voices run on the engine at the top of this page. Studio voices add emotion and stability controls. Tap play for a sample, or tap a free voice to load it into the generator.
Two clear, polite standard-Japanese voices you can use right now without an account.
Calm, measured voices with even pacing for long reads, documentaries and audiobook chapters.
Bright, natural voices for short videos, podcasts and casual scripts written in plain form.
Clear, unhurried voices that keep kanji-heavy sentences easy to follow at slower speeds.
Polished voices for keigo scripts — announcements, IVR menus, brand spots and product tours.
The studio adds 30 premium Japanese voices across narration, conversation, teaching and business. Open the premium Japanese voices
Use this voiceNo installs, no account. Kanji, hiragana, katakana and romaji brand names all go in the same box.
Drop your script into the editor at the top of the page. Use 。 and 、 as you normally would — the voice pauses on them, and a script with no punctuation runs together.
Names and place names are the usual traps. Write the kana in brackets right after the word — 東海林(しょうじ)— and the voice reads the kana instead of guessing.
Open the voice picker and choose by job: narration, conversation, teaching or business. A polite です・ます script sounds best on a narration or business voice; plain-form scripts suit the conversational ones.
Japanese narration runs around 300–350 characters a minute, which feels slower than English. Keep narration at 0.9–1.0×, push news reads to 1.1×, and drop to 0.8× for learners.
Press Generate, play the result, and hit download. The file lives only in your browser — nothing is uploaded to our servers.
Use the text to speech MP3 converter for longer passages, or open the full studio for premium voices and WAV export.
About 123 million people speak Japanese natively, and more than four million study it abroad — a record, according to the Japan Foundation's 2024 survey. Almost all of them read and write the same standard Japanese.
Japanese text to speech therefore lives or dies on register and pitch, not regional accent. Four things shape how a script sounds, and the text controls three of them.
「行きます」 and 「行く」 mean the same thing at different distances. A voice reads whichever you wrote, so match the form to the audience: polite for customers and narration, plain for vlogs, monologues and friends.
Japanese words are told apart by pitch, not stress. 雨 (rain) falls on the second mora, 飴 (candy) rises. Our voices use the Tokyo pattern that standard broadcast Japanese uses — the safest choice for any national audience.
Kansai speech flips several pitch patterns and swaps endings — 分からへん for 分からない, 〜や for 〜だ, おおきに for thanks. Every voice here speaks standard Japanese; you can write dialect words into a script, but the melody stays Tokyo.
Set phrases such as いらっしゃいませ and 〜でございます read naturally, which is why business voices work well for IVR menus and store announcements. Keep sentences short: keigo stacks quickly, and long clauses get hard to follow aloud.
| Scenario | Register | Voice type | Try |
|---|---|---|---|
| Store or phone announcements | Keigo (敬語) | Business | Kyoko, Makoto |
| Audiobooks, documentaries | Polite narration | Narration | Jin, Shiori |
| Short videos, vlogs | Plain form (だ) | Conversational | Akari |
| Courses, tutorials | Polite, slower | Educational | Satoru, Nanami |
A voice reads what you wrote, not what you meant — and in Japanese, one character can be read five ways. Before you run Japanese text to speech on a script, check these five things.
生 · 今日 · 東海林いきる / なま / せい · きょう · しょうじCommon kanji such as 生 and 行 carry a dozen readings, and the voice picks by context. Names and places are guessable at best — 東海林 is しょうじ, 日暮里 is にっぽり — so write the kana in brackets after any name you care about.
1本 · 3本 · 6本 · 4月4日いっぽん · さんぼん · ろっぽん · しがつ よっかCounters change the number's sound: 1本 is いっぽん, 3本 is さんぼん, 6本 is ろっぽん. Dates have their own set — 1日 is ついたち and 4月 is しがつ. Always write the counter after the digit so the voice knows which set to use.
はし箸 · 橋 · 端Chopsticks, bridge and edge all sound like はし and differ only by pitch. Write the kanji, not the kana, and the sentence around it tells the voice which pitch pattern to use.
おばさん · おばあさん · きて · きってaunt · grandmother · come · stampOne extra mora changes the word. Keep long vowels (ー, あ, う) and small っ exactly as written — a dropped character is the most common source of a wrong reading.
Google · Costco · IKEAグーグル · コストコ · イケアJapanese has its own readings for global brands, and they are not the English ones. Write brands in katakana when you want the Japanese reading, and leave them in Latin letters when you want them said in English.
Bracket the kana after names, write counters after digits, put brands in katakana, and end every sentence with 。. Four habits, and most misreadings disappear.
Four jobs people bring to this page every day.
Japan's audiobook market was projected at about ¥26 billion for 2024, up 86% on 2021. Use a narration voice at 0.9×, generate chapter by chapter, and keep each chapter under the character limit.
More than 11,000 Japanese YouTube channels have passed 100,000 subscribers. Voice the same script with a conversational voice for shorts and a narration voice for long-form, then drop the MP3 into CapCut or any editor.
Paste a passage, slow it to 0.8×, and read along — the voice resolves kanji you are unsure of. Pair it with a furigana tool and you have a listening drill for any text, not just textbook audio.
Japan's 2019 Reading Barrier-Free Act pushes publishers toward audio and accessible formats. Turn notices, menus and articles into clear standard-Japanese audio for readers who prefer to listen.
Everything on this page is free. Here is exactly what changes when you upgrade.
| Free | Premium voices | Your own voice | |
|---|---|---|---|
| Japanese voices | Free voices on this page | Studio voices by use, with emotion control | A clone of your voice, speaking Japanese |
| Characters per request | Up to 1,000 | Up to 5,000 | Up to 5,000 |
| Generations | 10 per day, free | Credits — 1 character = 1 credit | Credits — 1 character = 1 credit |
| Download | MP3 | MP3 and WAV | MP3 and WAV |
| Stored history | Nothing stored | Saved to your history | Saved to your history |
| Commercial use | Personal projects | Included with paid plans | Included with paid plans |
Previews, short clips, listening practice and quick social posts. Japanese characters count one each, so a free request covers a long paragraph.
Audiobooks, ads and courses where you want a specific register and steady delivery across many chapters. The studio keeps every generation for re-download. See plans and credits.
Upload a 15-second sample and clone your own voice. Your clone can read Japanese scripts even if you recorded the sample in English.
Small habits that make a generated voice sound like a person who meant it.
Japanese joins Spanish in our language pages. The free text to speech page covers every other language and accent in the catalog.
Free, no account, nothing stored. Paste your Japanese text, pick a voice by use, download the MP3.
Written by the AnyVoice team · August 25, 2026