Silence SpeedUp
Silence SpeedUp shortens videos by finding the silent stretches and either
speeding them up or cutting them out. FFmpeg does the encoding; this app decides
what to ask it for.
Made for lectures, interviews, podcasts and tutorials — anything where the
speaker pauses a lot or talks slowly.
Download
- Free, for Windows and Linux: from the
latest release.
The Windows installer is not signed, so Windows SmartScreen may warn about
an unknown publisher the first time; choose More info → Run anyway.
- From the Microsoft Store:
the same app, signed and kept up to date by the Store. It is paid, and
buying it is also a way to support the project.
- From source: see Building from source.
Coming from 2.x? See docs/porting-notes.md.
A presentation of the app, in English and Italian, is on
the author's site.
Getting started
- Add your videos — Add, a folder, or drag them onto the window.
- Adjust the settings if you want to; the defaults suit most recordings.
- Press the play button on a row to hear a short preview sample first,
then press Start.
Results land next to the original videos by default; another folder can be
chosen in the Export group of the encoding settings.
Full walkthrough of every setting: docs/usage.md.
Features
- ⏩ Speed up or remove the silences, and optionally mute them
- 🗣️ Speed up the speech too, if you want the whole video faster
- 🎙️ Tunable detection — noise threshold, minimum pause length, and a margin
so words are not clipped
- 📏 A measured threshold — the app reads the recording's background noise
and voice level and suggests the value between them
- ▶️ Preview — encode a short sample and hear the result before processing
the whole file
- 🔍 Quick analysis — every pause on a zoomable timeline, with the estimated
length of the result
- 🎧 Multi-track audio — keep every track while detecting the pauses from the
first one, as an OBS recording with voice and system audio needs
- 📋 Batch processing with a queue, a live log and a run you can stop
- 🎛️ Encoder controls — CRF, frame rate, x264 preset and tune, audio rate,
output container
- 🪟 Compact progress mode — a slim always-on-top strip for long batches
- 🌗 Light, dark or system theme, in English, Italian, German, Spanish,
French, Portuguese and Japanese
- 📦 FFmpeg included — nothing to install
Requirements
Nothing to install alongside it: FFmpeg ships with the app.
- Windows: 10 or newer, 64-bit.
- Linux: x86_64, with GTK 3 and
xdg-desktop-portal, which every full
desktop (GNOME, KDE, Xfce, ...) already has. Built and tested end to end
on Ubuntu 22.04. On Ubuntu 24.04, Debian 12 and Fedora 41 only its
libraries have been checked: they all load, but the app has not been run
there yet.
Building from source
Every platform needs Flutter 3.47 or newer, on the stable channel.
- Windows: Visual Studio 2022 with the Desktop development with C++
workload. The steps, and packaging the installer, are in
docs/development.md.
- Linux: clang, CMake, Ninja, GTK 3 headers,
patchelf, and the codec
libraries FFmpeg was linked against. The steps are in
docs/linux.md, which also explains how to build the Linux
version from Windows with Docker.
- macOS: Xcode and CocoaPods; then
flutter run -d macos. The macOS
build is not currently tested.
git clone https://github.com/padvincenzo/silence-speedup
cd silence-speedup
flutter pub get
flutter run -d windows # or: flutter run -d linux / -d macos
Documentation
Contributing
Welcome, in any of these forms:
- Reporting bugs
- Translating the app (copy
lib/l10n/arb/app_en.arb, change @@locale, and
run flutter gen-l10n; the new language appears in the menu on its own)
- Suggesting or implementing features from ROADMAP.md
- Buying me a coffee
Before a pull request: flutter analyze clean, flutter test green, and any
change to the FFmpeg calls covered by a test in
test/fragment_planner_test.dart or
test/preview_and_audio_test.dart.
Credits
AI assistance
Parts of the code were written with an AI assistant, then reviewed and
maintained by the author, who is responsible for it.
The app itself contains no AI, and sends nothing anywhere: the silences are
found by measuring the audio level with FFmpeg's silencedetect, which is a
threshold and a stopwatch, not a model.
Privacy
Videos are processed on your computer and never leave it. There are no
accounts, no analytics and no tracking. The only connection the app makes is
the update check of the GitHub version, which reads the public list of
releases; the Store version does not make it. The details are in the
privacy policy.
Licence
GPLv3 or later. Copyright (C) 2026 Vincenzo Padula.
This program comes with absolutely no warranty. It is free software, and you
are welcome to redistribute it under certain conditions; see LICENSE.
