← Hexwright Studios

Scribe — Offline AI Subtitle Generator and Speech to Text Software for Windows

Scribe — offline AI subtitle generator and transcription tool. 100% offline, 99 languages, GPU accelerated.

Drop a file. Get subtitles. Automatic subtitle generator with 99 languages, GPU accelerated, fully offline. Your files never leave your machine.

Get Scribe on itch.io $19.99 $15.99 launch discount

Features

Scribe is an offline speech-to-text tool that turns audio and video files into subtitles and transcripts, no internet connection needed.

Cloud transcription tools charge per minute. Scribe runs locally and you only pay once.

Drag & Drop

Drop any audio or video file: mp3, wav, mp4, mkv, webm, flac, and more.

Subtitles & Transcripts

Generate timed .srt subtitle files or plain .txt transcripts. Works as a video transcription tool or audio-to-text converter.

99 Languages

Automatic language detection. Supports 99 languages out of the box.

GPU Accelerated

Vulkan GPU acceleration for fast processing. Works on CPU too, just slower.

Fully Offline

One-time model download (~3 GB), then everything runs locally. No internet needed.

Audio Cleanup

Optional denoise and vocal isolation before transcription. Cleans up noisy audio for better results.

No Subscriptions

Buy it once, keep it forever. No accounts, no API keys, no per-minute charges.

Who's this for?

Why Scribe?

Most transcription tools are cloud services. That means uploading files, waiting for processing, paying per minute, and trusting someone else with your audio.

Scribe works differently:

Language Support

Powered by OpenAI's Whisper large-v3 speech recognition model, the same one used by professional transcription services, running entirely on your machine.

Excellent accuracy

English, Chinese, German, Spanish, French, Japanese, Portuguese, Russian, Korean, Italian, Dutch, Polish, Swedish, Turkish, Arabic

Good accuracy

Most European, East Asian, and South Asian languages

Variable accuracy

Less common languages may have reduced accuracy

Audio quality matters. Clean recordings (podcasts, lectures, interviews) give the best results. Heavy background music or overlapping speech will reduce accuracy.

System Requirements

Minimum

  • Windows 10 (64-bit)
  • 8 GB RAM
  • 4 GB storage
  • Any modern x64 CPU
  • 4 GB VRAM (GPU mode)

Recommended

  • 16 GB RAM
  • Vulkan-compatible GPU
  • 6+ GB VRAM
  • NVIDIA / AMD / Intel Arc

A GPU makes a huge difference in processing speed.

What's New

v1.2 — Audio Cleanup

Optional denoise and vocal isolation as a pre-processing step. Clean up noisy audio before transcription for better accuracy.

v1.1 — Batch Processing

Drop an entire folder. Scribe processes every file and saves subtitles alongside the originals. No more dragging files one at a time.

Read more on the devlog: v1.1 Batch Processing · Why I Built Scribe

What's Next

Scribe Software Roadmap — v1 Transcription (Now), v2 Translation & Audio (Next), v3 Extensions (Future)

This is AI transcription. It's good, but not perfect. Expect occasional errors with accents, mumbling, overlapping speakers, or noisy audio.

Some languages work much better than others.

Scribe is a transcription tool, not a translator. Translation is planned for v2.

Ready to try Scribe?

Buy it once, keep it forever. No subscriptions.

Get Scribe on itch.io $19.99 $15.99 launch discount