Download

Available on Windows & macOS

An all-in-one installer — every AI engine is built right in. Install once, run instantly, nothing else to set up.

🪟
Windows
Version 1.6.5 · Windows 10/11 (64-bit)
🧠 AI built in
.exe: fresh one-click install · 🔄 Update: for machines already installed — download the small package and run UPDATE.bat
🍎
macOS
Version 1.4.5 · macOS 12 or later
🧠 AI built in
.dmg file · pick the chip your Mac uses

Not sure which chip your Mac has? Click the Apple logo top-left → “About This Mac”. If it says “Apple M…”, pick the Apple-chip build; if it says “Intel”, pick the Intel build.

🚀 Why is this installer special?

Every AI model for listening, translating, generating voices and writing scripts is bundled and runs right on your machine — on both Windows and macOS. No external service account, no data sent to the cloud — fast and private.

System requirements
  • · Windows 10/11 (64-bit) or macOS 12+
  • · Mac: Apple chip (M1/M2/M3/M4) or Intel
  • · 8GB RAM or more (16GB recommended)
  • · ~10GB free disk space
  • · Discrete GPU (optional) for faster processing
Installation
  1. Download the installer that matches your machine.
  2. Run the file (.exe on Windows, .dmg on Mac).
  3. Open the app and copy your “Machine ID” to activate a license.
  4. Paste your license token (bought on the Pricing page).

✨ What’s new

Version 1.6.5
2026-09-02
  • New translation option: sign in with your Google account for very fast translation without an API key — a multi-hour film finishes in minutes (Professional and Lifetime plans)
  • Sign in with several accounts and rotate automatically when one runs out; enable or disable each account right in Settings
  • When Google asks an account to be verified, the app now says so and shows a verify button — it used to report a generic out-of-quota message so you waited for nothing
  • Redesigned Settings: fits on one screen, quota metrics at the top, rarely used sections collapsed
  • Fixed videos failing to render on films with many stretched lines — any number of lines now exports
  • Fixed Batch ignoring the export resolution you picked and always keeping the source size
  • The first-run AI model download window can now be moved and minimized instead of covering the screen
  • The missing-voice error now names the exact voice so you know what to fix
Version 1.6.4
2026-08-31
  • Speech extraction no longer invents lines nobody said over music or silence — it now uses a recognizer that does not keep "guessing ahead" when there is no speech, instead of only blocking known phrases
  • Blocks more repeated junk lines that slipped through at the end of extraction (fansub credits, a single character repeated for dozens of lines)
  • Translate tab: three new filter tabs — All caps+digits, All digits, Single word — to catch lines that used to slip past "Garbled/junk"
  • Translate tab: the Title panel and filter tabs no longer disappear when a video is stopped or fails partway
  • Translation: source lines misread into junk now get the real dialogue translated and the junk dropped, instead of translating the whole mess; genuine numbers are preserved
  • Translation: blocks made-up words that appear nowhere in the source
  • Framing: dragging the two video boxes now snaps to edges and centers automatically
  • Queue: videos that already have a translation skip the translation step and go straight to dubbing
  • Trimmed the text color palette down to the commonly used colors; fixed the blur layer being drawn over the wrong region in the preview
  • Voices now pronounce repeated-letter ranks correctly (S, SS, SSS, A, AA…)
Version 1.6.3
2026-08-29
  • The preview now matches the exported file exactly on fonts: it loads the real font files from your machine instead of guessing by name
  • Fonts that are not actually installed were removed from the list — picking one used to be silently swapped for another, making preview and export disagree
  • Bold is no longer faked on single-weight fonts; color band and title text sizes in the export now match the preview
  • Title box: type directly on the preview (double-click), break lines in place, align multi-line text and turn off the drop shadow
Version 1.6.2
2026-08-28
  • Dub tab: the "Keep original voice on overlong lines" option now keeps the original audio CONTINUOUS — most of it used to be dropped, so those lines sounded silent
  • The original audio on those lines is no longer clipped at the head/tail or uneven in level; lightly stretched lines let the original flow on without leaving choppy gaps
  • Dub tab configuration presets now capture EVERYTHING: text boxes, export resolution, reading speed, timing mode and the full audio mix — you no longer have to redo settings after loading a preset
  • CapCut voices read Vietnamese numbers correctly ("400" is no longer read as "tư trăm"), plus a substitution table for words this engine mispronounces (starting with "ừm")
  • Piper/Kokoro voices can now read common English words: added a ~90-word pronunciation table (dinosaurs, NASA, laser, vaccine, protein, sofa…)
  • Added a ChatGPT translation slot that shares all the rules and workflow with the Gemini slots; fixed films over 250 lines looping forever without translating
  • Fixed dub options being ignored when running through the Queue or Batch, and not being saved per project
Version 1.6.1
2026-08-21
  • Translate tab: replaced the "Review/Deep review" buttons with 4 quick filter tabs (Possible duplicate, Garbled/stray characters, Empty line, Untranslated) to jump straight to lines that need fixing, plus a button to re-translate the whole group
  • Catches more cases of stuttering ("kh kh...") or cut-off translations, and flags stray Korean/Arabic characters
  • Dub tab: added color grading & LUT for video, plus light subtitle entrance animation (fade/pop)
  • 10 built-in subtitle style presets (Netflix, TikTok, Cinematic, Period drama, News…) and configuration presets that save a full subtitle/logo/voice setup for reuse
  • Fixed the update (not fresh install) causing a black screen and immediate crash right after updating
Version 1.6.0
2026-08-16
  • Multi-voice dubbing in a single video: tag each character with a voice slot (Voice 1-5) and pick a different voice per slot — male, female, or a cloned voice
  • Automatic male/female voice detection during extraction, with quick manual override using number keys 1-5; voice tags now show on both the Extract and Translate tabs
  • Fixed subtitles being filled with lines nobody said (fansub credits, thank-you notes…) over music-only stretches
  • Fixed whole stretches of clearly audible dialogue going missing
  • Premium voices no longer add a stray word at the end of a sentence; about 35% faster on Windows machines with a dedicated graphics card
  • New voices: Lạc Phi (female), Thiện Tâm (male), Vũ Ngân (female)
  • Fixed the Translate tab freezing and refusing keyboard input while an export or background queue was running
  • Fixed the export resolution setting being ignored when exporting through the queue or in batch
Version 1.5.8
2026-08-10
  • Five new translation styles (Modern, Xianxia/fantasy, Historical drama, Natural, Tight dub — "new" variants) that keep forms of address far more consistent; the old styles are untouched so you can compare
  • Fixed music separation stopping midway on videos longer than 2 hours
  • Fixed a broken final-pass step that left many lines blank and untranslated
  • Fixed line misalignment when re-fixing flagged lines — one skipped line used to shift every line after it
  • Characters known only by their role (dad, mum, brother, sister…) with no proper name are now recognised — no more family members addressing each other like strangers
Version 1.5.7
2026-08-08
  • New background-music removal that keeps the original speech (Professional plan and up)
  • Pick your export resolution: 720p/1080p/2K/4K regardless of the source video
  • The home screen and preview open much faster when you have many projects
  • Fixed paid plans occasionally being detected as the free plan
Version 1.5.6
2026-08-06
  • Fixed the app losing its license after a few days — your activated key now stays put, no need to request a new one
  • License is checked daily but an active key is never revoked
  • Slow connections no longer corrupt translations: the app used to read Gemini replies before they finished and silently drop a whole batch
  • Subtitles read from video no longer appear ahead of the picture, especially on dialogue-heavy films
Version 1.5.5
2026-08-05
  • Added a "Modern (forced)" writing style — modern films no longer get translated into archaic period-drama pronouns
  • Much more accurate automatic detection of modern vs period films: 39/39 match across 39 real films
  • The "Re-export" button now works repeatedly — tweak subtitles, covers or signatures and export fast, without re-dubbing
  • Fixed losing subtitle text edits when switching projects or closing the app right after
  • Fixed one project translation bleeding into another when several projects are open
  • Preview subtitles no longer sit misaligned when you first open the Dub tab
  • Batch and background runs now use the exact voice and options you picked, instead of silently falling back to the default voice
  • Clear error when audio mixing drops lines, instead of reporting success on a video with missing speech
  • Merging vertical videos keeps the vertical frame instead of being forced to landscape
  • Cleans up leftover multi-GB temp files when dubbing fails
Version 1.5.4
2026-08-02
  • Fixed translated subtitles drifting out of sync with the dialogue — the app now detects and blanks suspect ranges so you can retranslate exactly there, instead of silently shipping wrong lines
  • Removed "ghost" subtitle lines that speech recognition invents (channel names, subscribe prompts) during passages with no speech
  • Fixed subtitles running out of sync when re-exporting a dubbed video in Stretch video / Hybrid mode
  • Fixed the "Smear blur" cover showing in preview but missing from the exported file
  • Translate tab no longer shakes or jumps around during extraction; scrolling and jump-to-line are much smoother
  • Blocked batches that come back as unreadable phonetic transliteration
  • Video stretching is more stable on subtitle-heavy films — no more mid-way failures
  • Removed the "long video, split it up" warning — the app now handles multi-hour videos in one pass
Version 1.5.3
2026-07-31
  • NEWA third dubbing mode — Hybrid: lines the voice can keep up with stay at their natural pace, only the long ones stretch the video. No breathless narration, no whole-film slowdown
  • NEWPick a writing style by genre: Historical, Wuxia, Xianxia, Game/system UI — characters address each other correctly from the start
  • NEWSimple translation mode: fewer constraints, letting the AI read each passage on its own. Good for films where roles are hard to infer
  • NEWA dedicated reader for offline voices: English words, abbreviations and symbols inside a sentence are pronounced, not spelled out
  • BETTERNoticeably clearer, smoother voices with no rasp when sped up
  • BETTERLarge numbers read naturally instead of as a long run of zeros
  • BETTERForms of address track the film better — historical and xianxia keep their classical register
  • BETTERSubtitles and watermark export exactly as they look in preview
  • BETTERMulti-hour films with tens of thousands of lines run smoothly — no freezing, no stopping midway
  • BETTERAfter an interruption, work resumes where it left off instead of regenerating every voice clip
Version 1.5.2
2026-07-28
  • Fixed extraction stopping unexpectedly on some videos (no more slow CPU fallback)
  • Faster, more reliable video muxing for long videos with many lines — fixes "Cannot merge video"
  • Much faster audio mixing on long videos (previously could crawl)
  • Titles generated after translation now follow the genre you picked
  • Chinese films: more natural pronouns when characters are unclear; "Concise dubbing" tone fits timing better
  • The interface no longer freezes when you switch to another window while translating or dubbing
  • The 3-day trial now unlocks all features with no limits
  • A character speaking several lines in a row now keeps one consistent way of addressing themselves
  • Finer scene-by-scene context splitting — pronouns and names now track the story more closely
  • Game-style rank/stat labels are now read out loud properly (e.g. "level 5") instead of misreading the abbreviation
  • Translate tab: now flags lines with an accidentally inserted speaker name or garbled text, with quick jump/clear buttons
  • New "Shared output folder" option in Settings — collects exported video/subtitles from every project in one place
  • OmniVoice dubbing now works right away on freshly installed machines (previously timed out)
  • Much faster first-time AI model download — the progress bar no longer throttles the transfer
  • RapidOCR text recognition now usable (previously silently fell back to another engine)
Version 1.5.1
2026-07-25
  • Correct number reading per language: numbers, dates, times, percentages, currency… read the way each language does (English, Chinese, Vietnamese…), no longer forced into Vietnamese
  • Content normalization when reading/translating per target language — smoother, more natural output for each language
  • Each voice reads in its own language; dubbing reads numbers in the translated language
  • "Translate again" button clearly re-translates from scratch — no longer shows the old translation while running
  • Re-translating a single line now works reliably (previously often failed on just one line)
Version 1.5.0
2026-07-19
  • Speech recognition now splits sentences at real pauses — no more first words sticking to the previous line
  • New online voice library: 129 voices in many languages, available in every dubbing tab
  • Added 7 new Vietnamese voices with better reading of numbers, dates and phone numbers
  • Star your favorite voices; unsupported voices are sorted to the bottom
  • API-key translation: auto-switches to backup models when your key lacks access + clear errors instead of leaving the original text
  • "Continue translating" button finishes the remaining lines after an interruption
  • Dubbing: a single missing clip no longer silences the whole video; duplicate dub runs are blocked
  • Voice generation on newest graphics cards auto-switches to a compatible mode — no more errors
  • Copy cover/blur boxes between projects; fixed watermark overlapping subtitles
  • Video merge: fixed broken frames/crackling audio with some formats
  • Background music playlist rotation; cleaner Dubbing and Batch UI; batch delete/resume mid-run
  • Complete English interface across the app
Version 1.4.3
2026-07-17
  • Added more natural-sounding online Vietnamese voices (risky cross-accent voices are flagged)
  • Star your favorite voices so they always appear at the top
  • More accurate speech-to-subtitles: fewer stray characters, no mid-sentence cuts, no dropped dialogue
  • Smoother video merging — no more broken picture/crackling audio on high-resolution videos
  • Dubbing & Batch: tidier interface with an intuitive voice grid picker
  • Batch: remove videos mid-run and "Continue" videos that already finished extraction + translation
  • Stronger voice protection
Version 1.4.1
2026-07-13
  • Batch: background music, remembers all settings (voice, scan, subtitles, cover boxes…), currency-conversion toggle
  • More reliable automatic video titles — fewer fallbacks to the default name
  • Ad-video insertion for Professional and Lifetime plans
  • Fixed subtitle box background color, video flip, and a Settings page crash
  • Better support for newer graphics cards, more stable
Version 1.4.0
2026-07-12
  • Free 3-day trial — activates automatically on first launch, no contact needed
  • Simpler activation screen with a one-tap "3-day trial" button
  • Stronger protection for license and voices
  • More stable overall
Version 1.3.13
2026-07-11
  • Translation is many times faster
  • Watch the translation appear live as it runs
  • If an API key runs out mid-way, it keeps going automatically
  • New "Continue translating" button to finish the remaining lines
  • Fixed missing/misaligned translated subtitle lines
  • Fixed the Minh Hoang voice on short sentences
  • Video merge: auto-renames on conflicts, remembers the save folder
Version 1.3.12
2026-07-09
  • NEW voice: Minh Hoang (male, custom-trained, offline)
  • Fixed subtitle line drift during translation + auto-detects and patches drifted lines when done
  • "Review translation" and "Deep review" buttons: AI scans & fixes translation errors (duplicates/omissions/drift/wrong meaning)
  • Mirror the video horizontally — see it instantly in preview, with text and logo not flipped
  • Batch tab: separate Signature and Logo (drag independently) + advanced options + tidier layout
  • New "Concise dubbing" style — translations that fit the speaking window
  • More even voice volume between sentences (no more loud/quiet swings)
  • Minh Hoang voice reads very short sentences better; premium voices no longer "time out"
  • Fixed jumpiness when deleting a line in the subtitle review screen
Version 1.3.11
2026-07-08
  • New-generation graphics cards (RTX 50xx): on-screen text reading runs correctly on the GPU
  • Machines without a discrete GPU still read text accurately (auto-switches mode)
  • Voice cloning runs entirely on your machine, no more network errors
  • Speech-to-subtitles: natural line breaks per sentence, noticeably denser lines
  • Tidier interface: removed redundant options, every mode is accurate
  • Updates only download the changes, keeping your existing data
Version 1.3.10
2026-07-08
  • Fixed batch dubbing stopping midway
  • New-gen graphics cards no longer freeze while reading on-screen text
  • Batch video: added a choice of Dubbing or Subtitles only
  • Watermark defaults to the center, freely draggable
Version 1.3.9
2026-07-08
  • Subtitle review is much smoother on Windows (no more lag clicking lines)
  • Faster, more stable translation that auto-patches error lines when done
  • Forms of address in the translation are more accurate per scene
Version 1.3.8
2026-07-08
  • Short subtitles translate much faster
  • Fixed a few untidy display spots
Version 1.3.7
2026-07-08
  • More stable text translation, less likely to be temporarily blocked
  • Freshly installed machines work right away, no more extraction errors
  • Fixed updates not applying on some machines
  • Hid some technical details for a cleaner look
  • Added the update history right inside the app
Version 1.3.5
2026-07-07
  • Update the app right inside it with one button
  • Installed machines only download the changes — very lightweight
Version 1.3.3
2026-07-06
  • Runs well on new-generation graphics cards
  • Reads Chinese subtitles more accurately

The installer bundles the full AI engine, so it is large. An Internet connection is needed once for activation.

fZ