v0.2 · early development

Local subtitles for your own broadcasts.

Scrybe runs local Speech-To-Text models on your machine to create live captions that you can add to OBS as a browser source. Free, Fully local, open-source, no cloud. Still in early development, expect lots of changes.

Download for macOS
v0.2.16 · AGPL-3.0 · All releases
Scrybe overlay editor
and this is what it looks like on stream
What it does

Runs locally

All processing happens on your machine. Your audio never leaves your computer. This means no API keys, no cloud latency, no subscriptions.

Mic or desktop audio

Select any audio device. Caption dekstop audio, your voice through your microphone, even virtual audio cables.

OBS browser source

Drop localhost:3030 into OBS and the subtitles appear. Adjust text alignment, color, font size, transparency, etc and the overlay will update automatically.

Advanced Settings

Swap models, toggle translate, trim silence, etc. Works great out of the box, but allows you to customize your experience.

Planned features

What is coming next.

Scrybe is still early, and the roadmap is focused on making live captions more flexible, readable, and useful after the stream.

Smarter audio capture

Choose the exact audio source, whether that's a microphone, desktop audio, or audio from a specific process. Follow the system default when it changes, and keep transcription running when it switches.

Cleaner live transcripts

More natural caption segments with silence and speaker-aware breaks, steadier updates, accurate timestamps, and speaker identification.

More OBS overlay options

A segment-log overlay, saved custom overlay colors, and dedicated OBS plugin support for Twitch native captions and microphone mute events.

More control over transcripts

Save transcripts to disk, dedicated translation support, and transcript history.

Setup

Four steps from install to on-stream subtitles.

Scrybe is a desktop app based on Tauri, install it like any normal program. Choose from a set of model presets (or bring your own) and start transcribing.

  1. 1 Install and launch Scrybe.
  2. 2 Customize the subtitle box's positioning, color, and size.
  3. 3 Add http://localhost:3030/app/v1/overlay to OBS 1920 × 1080
  4. 4 Press Start and captions appear. go live
under the hood

Powered by whisper.cpp.

C/C++
no runtime, no Python
CPU + GPU
Metal (Apple), CUDA (Nvidia), Vulkan (cross-platform)
$0
no API bills, no account creation

whisper.cpp is Georgi Gerganov's port of OpenAI's Whisper speech-recognition model to plain C/C++. Same models, same weights, it's just stripped of the Python stack and compiled down to something small enough to ship inside a desktop app.

Scrybe embeds it directly. Audio goes in, text comes out, nothing is sent remotely. You pick the model size tiny for a low-end laptops, large-v3-turbo for when you want more accuracy.

More at github.com/ggml-org/whisper.cpp.