Turn Japanese speech into romaji subtitles. AI transcribes the Japanese and writes it in Latin letters, with the kana line underneath if you want both.
Free for recordings up to 1 hour and 1 GB. Longer files need a paid plan.
How It Works
Drop the Japanese audio or video you want rewritten.
The speech is transcribed, then every line is written in Romaji (Latin-script Japanese).
Download the SRT or VTT, or copy the text — the original-language file comes with it.
The Romaji Subtitles is a free online tool from ShortsHero: japanese speech in Latin letters. The speech is transcribed by AssemblyAI, then every line is translated with its timings kept — usually under a minute for a 5-minute file. It is free for files up to 1 hour; longer recordings need a paid plan.
Japanese is in AssemblyAI's high-accuracy tier, with a typical word error rate of 10% or less on clear speech. Japanese also runs on AssemblyAI's newest Universal-3.5 Pro model, which handles regional accents and mixed-language speech.
Free transcription covers files up to 1 hour. If your recording is longer, a paid plan lifts the limit — or trim it into parts and run each one.
Why Use This Tool
Line by line, so every subtitle keeps the timing of the speech it came from.
You get the source-language subtitles too, to check or reuse.
Ready for YouTube, Vimeo, Premiere, Resolve and every major player.
Whole recordings, not a sample.
Use the files anywhere — the transcript and subtitles are yours.
Open the SRT in our free subtitle editor to correct names and terms.
Yes. Files up to 1 hour long are free, with no watermark and no account needed. Recordings longer than 1 hour need a paid ShortsHero plan.
Japanese is in AssemblyAI's high-accuracy tier, with a typical word error rate of 10% or less on clear speech. Japanese also runs on AssemblyAI's newest Universal-3.5 Pro model, which handles regional accents and mixed-language speech. Accuracy drops with music, crosstalk and distant microphones, so check names and technical terms before you publish.
MP4, MOV, WebM, MKV, MP3, WAV, M4A, AAC, OGG, FLAC and most other audio and video formats, up to 1 GB. For a long video, uploading just the audio is faster.
No — the words stay in the original language and are written in Romaji (Latin-script Japanese), so you can read along without knowing the original script.
Usually well under the length of the recording — a 10-minute file is typically done in about a minute.
ShortsHero turns long videos into short vertical clips with AI moment detection, face tracking and animated captions — and has free tools for subtitles, transcripts and video formats.
The same transcription powers ShortsHero's clip maker: it finds the strongest moments, reframes them to vertical and adds word-by-word captions in 36 styles.
Try the clip maker freeFree to start · No credit card required
Keep Going
Make romanized Korean subtitles. AI transcribes the Korean and writes it in Latin letters so fans can sing along, with Hangul and English lines optional.
Add pinyin subtitles to Mandarin video. AI transcribes the Chinese and shows tone-marked pinyin under each line, made for learners and lyric videos.
Turn Arabic speech into Arabizi (Franco-Arabic) subtitles written in Latin letters and numbers, the way people text, with Arabic script optional.
Turn any video into a vocabulary list. AI pulls the useful words and phrases, adds translations and example lines with timestamps, ready for flashcards.
Practise shadowing with any video. It plays one sentence, pauses for you to repeat, then moves on, with the text on screen and adjustable speed.
Check a video's language level from A1 to C2. AI reviews the vocabulary, speed and grammar used so learners can pick videos that fit their level.