Mac · Apple silicon · Mac App Store
LoudnessKit
Automatic audio mastering, with repair for AI-generated tracks
LoudnessKit masters a track automatically. It measures it, identifies the genre, fixes the typical weaknesses of AI-generated music and brings it to the loudness of the platforms you choose. Then you listen, compare with the original and adjust what you want to hear differently. Everything runs locally on your Mac.
For Macs with Apple silicon and macOS 15 or later.
Coming soon to the Mac App Store.
Why LoudnessKit?
Three core arguments
Automatic, then yours
LoudnessKit measures the track, estimates the genre and computes a master right away. Every slider starts on Auto; move one and a new preview follows after a short pause.
Made for AI-generated music
Noise-like highs, harshness around 5 kHz, mud in the low mids and a high-frequency cutoff from MP3 delivery are diagnosed and repaired, with a strength derived from the analysis.
Runs on your Mac
Analysis, mastering and export run locally. No account, no cloud, no data collection.
Features
What’s inside
Analysis
Loudness (LUFS), true peak, loudness range (LRA), peak-to-loudness ratio, stereo correlation and mono loss, each for the master and the original. A high-frequency diagnosis tells a cutoff, an early roll-off, noise-like highs and a healthy top end apart.
Genre detection
The genre is estimated from the sound profile, tempo and detected instruments. If the detection is uncertain, LoudnessKit says so, and you pick the genre yourself. It sets the target curve, loudness, dynamics, width and space.
Character and tone sliders
Brilliance, warmth, punch, transparency, space and width, from −100 to +100 with Auto in the center. A character such as Warm or Brilliant sets the starting point.
AI repair
De-shimmer, de-fizz and de-mud tame noise-like highs, harshness and muddiness. Bandwidth extension restores highs above a detected cutoff. Automatic by default, or set each component by hand.
Space and width without mud
The artificial space acts only above 300 Hz, steps back while the music is dense and sits mainly on the sides. Width works on the side signal; the bass below 120 Hz stays mono, and the mono guard pulls the width back if a mono fold-down would lose more than 3 dB.
Output targets
Streaming, Spotify, Apple Music, YouTube, CD, broadcast (EBU R128) and more, several in one pass. Every finished file is measured again for loudness and true peak; if the peak exceeds the limit, the file is rendered again with more headroom.
Compare and see
Switch between original and master with B, loudness-matched, so you judge the sound and not the level. Spectrogram before and after, loudness over time, long-term spectrum and stereo image.
Presets and projects
Presets keep genre, character, sliders, repair and targets for a series of tracks. A project (.loudnesskit) keeps the audio file together with all settings.
Help in 18 languages
The built-in help explains every step, every value and every keyboard shortcut. Open it with Help ▸ LoudnessKit Help or ⌘?.
Control by AI models
A built-in MCP server lets AI models such as Claude analyze and master tracks or whole folders. It listens only on this Mac, is off by default and requires a token. Shortcuts actions “Master Track” and “Analyze Track” work without it.
Screenshots
The app at work
Screenshots in dark mode. Click to enlarge.
Demo file for testing
Try it with an AI-generated track
A 45-second synthwave excerpt, generated with an AI music generator, as a 16-bit stereo WAV at 44.1 kHz. Open it in LoudnessKit and look at the diagnosis, the repair and the before-and-after spectrogram.
Download demo file (WAV, 7.9 MB)Support
Help, bug reports and ideas
The complete manual is built into the app: Help ▸ LoudnessKit Help or ⌘?. It covers the analysis, the sliders, AI repair, output targets, AI control and every keyboard shortcut.
For bug reports and feature requests, please send an email with the subject “LoudnessKit Feedback”. The app version, your macOS version and a short description of the steps help me most. I read and answer every message myself.
- info@kindler-dev.de
- Phone
- +49 89 62084444
- Developer
- Björn Kindler, Kindler Dev, Bischofshofener Str. 9, 82008 Unterhaching, Germany
System requirements
- Mac with Apple silicon
- macOS 15 or later
- Input: WAV, AIFF, FLAC, MP3 and AAC, mono or stereo, from 32 to 192 kHz
- No account and no internet connection needed for mastering
FAQ
Short answers
Does my audio leave the Mac?
No. Analysis, mastering and export run locally. LoudnessKit has no account, no cloud processing and no analytics.
Why does an export stay quieter than its target?
On purpose. The limiter may act by no more than 6 dB at peaks and 3 dB on average, otherwise attacks and dynamics would suffer. Loud targets such as Club or CD can then stay below their value, and the export report says why.
The genre is wrong. What now?
Pick it yourself in the sidebar. LoudnessKit points out when the detection is uncertain, and the genre decides the target curve, loudness, dynamics, width and space.
Is LoudnessKit only for AI-generated music?
No. Analysis, tone sliders, space, width and output targets work for any track. The AI repair only acts as strongly as the diagnosis calls for, and you can turn it down or set it by hand.
How do AI models control the app?
Turn on the MCP server under Help ▸ Set Up AI Control and run the command shown there once in Terminal. The server listens only on this Mac, requires a token and may only read and write in folders you share.
Privacy
Your tracks stay on your Mac.
LoudnessKit collects no personal data and makes no outgoing network connections. No account, no analytics, no tracking. The optional MCP server for AI control is off by default and only reachable on this Mac. Full details in the dedicated LoudnessKit privacy policy.
Questions? Get in touch