Windows app · Offline Whisper

Subtitles from any audio or video.
On your own computer.

Drop an MP3, a song or a movie. MAKAN Media Translate listens with offline Whisper, writes timed subtitles, translates them, and saves files your player understands. Your files never leave your PC.

  • Windows 10 & 11
  • Files stay on your PC
  • English · فارسی

Four steps

How it works

  1. 01

    Drop your file

    MP3, WAV, M4A, FLAC, MP4, MKV and more.

  2. 02

    Speech to text

    Whisper runs on your PC and can use your graphics card.

  3. 03

    Translate

    Pick a language. Offline engines first, with automatic fallback if one stops.

  4. 04

    Preview and save

    Play with live subtitles, fix a word, choose font, size and colour, then save.

Watch

See it in action

Two short guides recorded in the real app.

Quick startA short tour: drop a file, transcribe, translate and save.
Full guideA complete walkthrough of MAKAN Media Translate, step by step, using a real song.

Features

Features

Offline audio to subtitles, translation and styled subtitle files in one Windows app.

Offline speech to text

Whisper works without internet after the first model download.

Optional GPU speed-up

Uses your graphics card when it can, with automatic fallback to the processor.

Many engines, one ordered chain

Local models, Ollama (for example TranslateGemma), LM Studio-style servers, Chrome’s on-device translator, and optional free web translators (off until you allow them) and Gemini with your own key.

Live preview

Click a line to jump, edit text, and switch between original, translated and bilingual views.

Subtitle style

English and Persian fonts, size, 24 colours or your own colour, and a background box. The chosen style can be written into the files.

Save SRT, VTT, ASS or text

Save as SRT, VTT, ASS (styled) or plain text, next to your media file in one click.

Made for songs too

Lyrics appear in a caption box under the player. ASS files show styled lyrics on MP3 in PotPlayer.

Built-in User guide

A User guide in English and Persian inside the app.

Installer or portable

A Windows installer with uninstaller, or a single portable file.

Formats

Input: MP3, WAV, M4A, AAC, OGG, OPUS, FLAC, MP4, MKV, WEBM, MOV, AVI

Output: SRT, VTT, ASS, TXT

MKV, AVI and very long videos need the free ffmpeg (optional, not included). ffmpeg.org ↗

System requirements

  • Windows 10 or 11, 64-bit.
  • 8 GB RAM minimum, 16 GB or more recommended.
  • A recent NVIDIA/AMD graphics card helps but is not required.
  • Free disk space for models (a few hundred MB to several GB).
  • Internet only for downloading models.

Screenshots

Screenshots

the first page with the empty drop zone
lines produced from a song or video
original and translated lines
the preview with the Subtitle style panel open
the Model Hub with the engine cards and the fallback chain
the built-in User guide page

Download

Download

For Windows 10 and 11. Compare the SHA-256 with Get-FileHash in PowerShell to check your download.

Installer

MAKAN-Media-Translate-Setup-0.1.0.exe

Installer78.0 MB
SHA-25625093542eaf23559ba75eb23d8fa5c8d45406d589def5f79610e2693cb9f1797

Portable

MAKAN-Media-Translate-portable.exe

Portable70.9 MB
SHA-256d241b68c103be3e9b2e1507ce8d5c83a4e5584fb2dafcaa9d8c05dcc1d607fb4

The installer is not code-signed yet. If Windows shows “protected your PC”, click More info, then Run anyway. ffmpeg is not included; it is optional.

Privacy

Privacy

Speech recognition and offline translation run on your computer. Your files are not uploaded. Only the optional free web translators (if you switch them on) and Gemini (if you add your own key) send subtitle text over the internet. Keys are kept in memory only and are never saved or written into files.

Honest limits

Honest limits

  • Whisper can miss or invent words, especially in songs. Persian singing is the weakest case.
  • Translation quality depends on the engine. The best results we saw came from a local TranslateGemma model through Ollama.
  • Free web translators are unofficial and may stop working.
  • The first model download needs internet.

Credits and licences

Credits and licences

  • Whisper (OpenAI) - MIT licence. Run through Transformers.js (Apache-2.0).
  • NLLB-200 (CC BY-NC 4.0), M2M-100 and mBART-50 - NLLB is for non-commercial use only.
  • Electron and electron-builder - MIT.
  • ffmpeg - not included; you use your own copy.
  • Full list: docs/THIRD_PARTY.md in the app folder.

FAQ

FAQ

Does it work offline?

Yes for speech to text and for downloaded translation models. The first model download needs internet.

Which player shows my subtitles?

Any player that loads SRT, VTT or ASS files. For MP3 files PotPlayer works great with ASS.

Why are some song lyrics wrong?

Singing is harder than speech. Check and fix lines in the preview.

Windows says “protected your PC”. Is that normal?

The installer is not code-signed yet. Click More info, then Run anyway.

Does it burn subtitles into video?

No. It makes subtitle files. Use your player or a video editor for burning.

Is ffmpeg included?

No, because of its licence. It is optional and you use your own copy from ffmpeg.org. MKV, AVI and very long videos need it.