From a janky screen recording to a perfectly shareable video asset in a single step.
Just drag your file onto Louder to enhance loudness, crop silence, remove noise, and encode — all at once.
See the project page for the changelog and screenshots.
Screencasts, demos, and meeting recordings almost always share one problem: the picture is fine, but the voice is quiet, thin, and buried under room noise.
Pick a preset once and it writes the result back in place. No timeline, no plugins, nothing to configure.
- Just drop it — Drag any video onto the window or the Dock icon and Louder takes it from there — no import dialogs, no settings to wade through. Your file is replaced with the enhanced version in a single step.
- Built for EarPods — Screencasts recorded on EarPods or a laptop mic come out thin, quiet, and easy to miss. Louder is tuned for exactly that voice — turning a janky take into something you'd actually share.
- Compare before you commit — Not sure which preset to trust? Compare mode renders them side by side so you can hear the difference for yourself before you pick the keeper.
- See what changed — A clear visual readout — loudness curves, integrated LUFS, and estimated SNR — shows exactly how much quieter the noise got and how much louder the voice came through.
Get the latest build from the Releases page, unzip, and drag Louder.app into Applications. Apple Silicon only; you still need ffmpeg (see Requirements).
The app is signed with a Developer ID and notarized by Apple, so it launches without Gatekeeper warnings. Prefer to build it yourself? See Building below.
Drag one or more video or audio files onto the Dock icon or the drop window. For each file, Louder:
- Backs up the original to
<name> - original.<ext>beside it. - Applies one remembered preset — Louder, Studio, Focus, or Clean (see Presets) — targeting a consistent −16 LUFS.
- Adds natural 0.25-second fades at the start and end (toggle in Settings), then re-encodes audio as broadly-compatible 48 kHz AAC-LC with
+faststart. - Replaces the original in place. If a file fails, the original is left untouched.
Pick Compare (the last menu item) to leave the source alone and write one clearly named variant for every preset beside it, so you can A/B them. Each variant is saved as <name> - <Preset>.<ext> (e.g. talk - Studio.mp4).
After processing, the window stays open with compact loudness curves, integrated LUFS, an estimated signal-to-noise ratio, and a mode-aware Reset. Click any curve to hear that version; switching keeps the current timestamp for instant A/B. Reset returns you to the drop screen — in Compare it clears every generated variant; on a single file it keeps the louder result and asks whether to keep or delete the original.
Pick one once. Each cleans the voice a little differently, then brings it to a consistent −16 LUFS (EBU R128), encoded as 48 kHz AAC-LC. The ⓘ button beside the menu opens a schematic signal chain showing the exact models and parameters behind the selected preset.
| Preset | Signal chain | What it's for |
|---|---|---|
| Louder | Denoise → Loudness | The safe default — gentle cleanup, then a consistent level. DeepFilterNet3 removes steady background noise and hiss, keeping the voice intact. |
| Studio | Denoise → EQ → Compress → Loudness | A warm broadcast tone — rumble cut, presence shaping, gentle compression. |
| Focus | Gate → Denoise → Loudness | Best for occasionally noisy rooms — a door, a dog, keyboard clatter. Ducks intermittent events between words (Apple SoundAnalysis) before denoising. |
| Clean | Enhance → Loudness | The strongest cleanup — try it when the others leave too much noise. Rebuilds the voice with an on-device neural model (GTCRN); band-limited, so clearer but slightly duller. |
A small 4K / 1080p control on the drop screen sets a maximum output height (2160 or 1080) and is remembered between launches. It's a cap, not a forced size:
- A recording at or below the cap keeps its original video, copied untouched (the default, fastest path).
- A recording above the cap is downscaled and re-encoded to broadly-compatible H.264 8-bit 4:2:0 (CRF 18,
veryfast), preserving aspect ratio.
Whenever the video is actually re-encoded — by a downscale or a silence trim — a Size card shows the file-size change (e.g. 120 MB → 66 MB); in a Compare batch the smallest variant is starred.
Output audio is always 48 kHz AAC-LC and every file is written +faststart for instant playback on the web and devices. The video stream is copied untouched to preserve quality and speed, so its compatibility matches your source — Louder inspects the result and warns about limitations like H.265 (HEVC), high-bit-depth/4:2:2 pixel formats, or an audio/video length mismatch. For the broadest reach, record in H.264.
Mono and stereo audio pass through untouched; surround (more than two channels, e.g. 5.1) is merged down to mono so it can't play back silent or broken on devices that can't decode the layout.
- macOS on Apple Silicon (the bundled DeepFilterNet denoiser is an
arm64binary). - ffmpeg via Homebrew:
brew install ffmpeg. Louder looks in/opt/homebrew/binand/usr/local/bin. If ffmpeg (or Homebrew) is missing, the app shows a guided setup card on launch with copy-ready install commands, an Open Terminal button, and Re-check — no need to drop a file first.
DeepFilterNet 0.5.6 and the DeepFilterNet3 model are bundled. All cleanup runs locally; recordings are never uploaded.
Open Louder.xcodeproj in Xcode and run — nothing else to configure. The app is deliberately not sandboxed so it can rewrite files in place anywhere on disk, including cloud-synced folders like OneDrive (which triggers a one-time macOS permission prompt).
A Shortcuts Quick Action was tried first and abandoned: Quick Actions run inside Apple's sandboxed siriactionsd daemon, which can never be granted access to OneDrive/CloudStorage files. A real app in the user session gets the prompt once and works forever.
No editing, format conversion, folder watching, arbitrary audio parameters, or App Store distribution. It's a personal tool that does one thing well.
Louder builds on the work of these projects — and the people behind them: FFmpeg, DeepFilterNet, GTCRN, EBU R128, Inconsolata, Lucide, and Swift & SwiftUI.
Licensed under PolyForm Noncommercial 1.0.0 — free to use, modify, and share for noncommercial purposes.
Code by Bernd Plontsch and Claude Opus 4.8.