00:00:01,000 → 00:00:04,200 · SRT · 24 fps

Every film deserves
perfect subtitles.

KenjaSub is an AI subtitle editor for Mac. Drop in any video — transcription runs entirely on your machine, translation happens in context with the AI model of your choice, and a frame-accurate studio polishes every line. SRT & WebVTT out.

Download on the Mac App Store · Free GitHub · KenjaHub macOS 26+ · Apple Silicon recommended
KenjaSub subtitle studio on Mac: video player, synchronized cue list, waveform and AI assistant panel
REFINE
TC 00:00:24:12
転写も翻訳も、この Mac の中だけで完結する Transcription and translation never leave your Mac 转写与翻译,全部在本机完成 — KenjaSub, the sage's subtitle studio
01

Three tabs, one workflow

From raw video to shipped bilingual subtitles without leaving the app — batch it, refine it, and bring your own model.

TAB 01 · AUTO

Batch, hands-off

Drop several videos and get subtitle files. Speech recognition, translation and export run as one queue — per-video overrides, live progress, one-click retry.

KenjaSub Auto tab: batch transcription and translation queue for multiple videos
TAB 02 · REFINE

The subtitle studio

An mpv-based player opens MKV, WebM and AVI natively. Cue list and waveform stay in sync; drag timings, split and merge cues, shift the whole track, undo anything, export SRT or WebVTT.

KenjaSub Refine studio: bilingual cue list, waveform timing and video preview
TAB 03 · MODELS

Bring your own key

Any OpenAI-compatible provider — OpenAI, DeepSeek, OpenRouter, Ollama, LM Studio or your own endpoint. Keys live in your Mac's Keychain, never in plain files.

KenjaSub Models tab: manage OpenAI-compatible AI providers with keys stored in Keychain
02

Built for people who care about lines

Not a cloud dashboard — a native Mac tool with an opinion about quality.

TRANSCRIPTION

On-device, always

Speech recognition runs with Apple's SpeechAnalyzer. Your audio never leaves the Mac — no uploads, no queue on someone else's server.

TRANSLATION

Reads the whole film

Lines are translated in context windows, so pronouns, tone and recurring names stay consistent from opening scene to credits. Six registers, from Literal to YouTube.

ASSISTANT

Chat with your subtitles

Retranslate a scene, shorten a line, fix a name — conversationally. "Ask before changes" stages every edit as a reviewable diff; automatic mode applies instantly with one-click undo.

TIMING

Frame-accurate control

Drag cue boundaries on the waveform, set in/out points at the playhead, type exact timecodes, split, merge, or shift the entire track. Full undo history included.

FORMATS

MKV in, SRT/VTT out

Native MKV, WebM and AVI playback via mpv. Bilingual SRT imports with automatic language detection — your existing translations come along.

QUALITY

Readable by design

Reading-speed and line-length checks keep subtitles comfortable on screen. When a source language or model answer looks wrong, KenjaSub stops and tells you.

03

Private by default

No accounts. No analytics. No tracking.

KenjaSub is private the way a hammer is private — it's a tool on your machine, not a service with a login. The rare data that moves, moves only when you choose an engine and only to the provider you configured.

Read the privacy policy →

✓Transcription: 100% on device
✓Local engine option: Apple Translation, on device
✓API keys: stored in Keychain only
✓Telemetry: none, ever
04

Questions, answered

Does KenjaSub upload my video or audio anywhere?

No. Speech transcription runs entirely on your Mac with Apple's SpeechAnalyzer — audio never leaves the device. When you choose an AI translation engine, subtitle text is sent only to the provider you configured, only when you run that engine.

Which AI models can it use for translation?

Any OpenAI-compatible provider: OpenAI, DeepSeek, OpenRouter, Ollama, LM Studio or your own endpoint — keys live in your Mac's Keychain. A fully local option using Apple's Translation framework is built in as well.

Which video and subtitle formats are supported?

The player opens MKV, WebM and AVI natively (mpv-based), plus standard MP4 and MOV. Subtitles import and export as SRT and WebVTT, including bilingual SRT files with automatic language detection.

Can I edit subtitle timings manually?

Yes — drag boundaries on the waveform, set in/out points at the playhead (⌘[ / ⌘]), type exact timecodes in the edit sheet, split and merge cues, shift the whole track earlier or later, and undo any of it.

What are the system requirements?

macOS 26 or later. On-device transcription and translation use Apple's neural frameworks, so Apple Silicon is recommended.

Does KenjaSub include AI model access?

No — AI features are strictly bring-your-own-key. You add a provider and key of your choice; KenjaSub never bundles or resells model access, which keeps your costs and your data under your control.