Darwin's Cat

Professional Automatic Mastering Service

EN DE FR PL CS IT ES UK RU

0:00 / 0:00

How to use

  1. Drag and drop a WAV file into the zone (or click to select). The tool measures loudness (LUFS), true peak, dynamic range (LRA), sample rate, bit depth and duration. Metrics appear within a couple of seconds, with the waveform and player below them.

  2. Pick Platform (Streaming, YouTube or CD), Style (Jazz, Hip-Hop, Electronic, Pop, Country, Rock, Metal), Density (Light, Medium, Heavy), Dynamics (Dynamic, Balanced, Loud, Extreme) and, if you know it, the track BPM. The algorithm recomputes the targets automatically.

  3. Hit Start. The indicator at the top shows progress through stages (analyzing β†’ compressing β†’ limiting β†’ verifying). The finished track appears below β€” with its own waveform, player and download button.

You can change parameters and run mastering again without re-loading the file. Each run is saved as a separate block β€” download any of them.

Advanced

The collapsible Advanced block lets you override any parameter manually:

  • Sample Rate of the output (44.1 / 48 / 88.2 / 96 kHz, or keep input)

  • Limiter: Target LUFS, True Peak ceiling, oversampling (1 / 2 / 4 / 8Γ—)

  • EQ: HPF frequency and the gain of five bands (Low End, Warmth, Mud, Brightness, Air)

  • Compressor: Ratio, Thresh Offset, Attack, Release, Knee

Any of the three blocks (Limiter / EQ / Compressor) can be fully disabled with a checkbox.

Limiter

The limiter is the main mastering tool. It raises the overall track loudness to the target level and prevents peaks from exceeding the allowed ceiling.

Why is this needed?

Streaming platforms normalize loudness β€” they’ll turn up a quiet track and turn down a loud one. But if peaks are too high, distortion will appear during MP3/AAC conversion. The limiter solves both problems: makes the track loud enough and protects against clipping.

Iterative precision:

The limiter works iteratively β€” up to 8 attempts until it hits the target within Β±0.2 dB. After each attempt, the actual LUFS is measured, and gain is corrected. This guarantees precise target hitting even with aggressive settings.

If the target is physically unreachable (too loud for the given material) β€” the algorithm detects this and stops, delivering the best possible result.

Parameters:

  • Target LUFS β€” target loudness. Calculated automatically: platform + style + dynamics.

  • True Peak β€” peak ceiling. -1.0 dB for streaming, -0.3 dB for CD. A safety margin guarantees non-exceedance.

  • Oversample β€” processing precision. 4Γ— or 8Γ— for reliable inter-sample peak control. Signal is upsampled, limited, then downsampled to the final sample rate.

Compressor

The glue compressor binds the mix together, adds punch, and controls dynamics. It works gently β€” doesn’t kill the liveliness, but makes the sound tighter and more cohesive.

When is it needed?

For most modern genres β€” Rock, Pop, Electronic, Hip-Hop, Metal. For Jazz and classical music, you can disable it to preserve natural dynamics.

Smart threshold:

The threshold is calculated not from peaks (which are random), but from the loudest sections of the track β€” 95th percentile of short-term LUFS (3-second windows). This means the compressor responds to actual music β€” choruses, drops β€” not a random stick click or clap.

Parameters:

  • Ratio β€” compression amount. 1.5:1 β€” gentle glue, 4:1 β€” tight control. Calculated from Density + Dynamics.

  • Thresh Offset β€” activation-threshold offset in dB, relative to the loudest sections of the track. Higher offset = higher threshold = less compression.

  • Attack β€” reaction speed. Fast (8–12 ms) for transient control, slow (30+ ms) to let drum clicks through. Metal and Electronic get fast attack, Jazz and Country get slow.

  • Release β€” recovery time. Calculated from BPM: 60000 / BPM / divisor. Faster tempo = shorter release. The compressor breathes in rhythm with the music.

  • Knee β€” softness of onset. Hard (1–2 dB) for punch and aggression, soft (6–8 dB) for transparent gluing. Metal gets hard knee, Jazz gets maximum softness.

Smart automation:

All parameters adapt to the combination of style, density, and dynamics. Metal + Heavy + Loud = fast attack, hard knee, aggressive ratio. Jazz + Light + Dynamic = slow attack, soft knee, minimal compression (or complete bypass). Everything can be overridden manually in Advanced mode.

Equalizer

Tonal correction for final polishing. The high-pass filter removes unwanted rumble, and five bands let you adjust the tonal balance.

HPF (High-Pass Filter)

Cuts sub-bass garbage below 15-50 Hz. The cutoff frequency is automatically selected based on genre and density: lower for Electronic and Hip-Hop (15 Hz β€” infrasound only), higher for Rock and Metal (30-35 Hz β€” removes rumble). Heavy Density adds +5 Hz, Loud Dynamics adds +6 Hz. Works before the compressor β€” so the compressor doesn’t react to sub-bass garbage.

Tonal bands:

Band Frequency Purpose When to use

Low End

80 Hz (shelf)

Adds weight and sub-bass

Mix sounds thin/empty, lacking foundation. Don’t use on already bassy mixes.

Warmth

150 Hz (bell, Q=0.7)

Body and fullness

Mix sounds cold/thin in the low-mids. Adds "meat" to guitars and vocals.

Mud

290 Hz (bell, Q=0.8)

Removes muddiness (cut only)

Mix sounds muddy/boxy. Common problem with home recordings. Cuts -1.5 or -3 dB.

Brightness

8 kHz (shelf)

Presence and clarity

Mix sounds dull, vocals get lost. Adds "sparkle" and intelligibility.

Air

12 kHz (shelf)

Air and openness of highs

Want a sense of space and "expensive" sound. Careful β€” may emphasize sibilance.

Each band has several fixed values β€” from off to +3 dB. Simple choice instead of endless knob turning.

Tip
 If you’re not sure you need EQ β€” don’t enable it. Mastering EQ should correct, not color. If the mix is well balanced, EQ isn’t needed.

Platforms and Formats

Each platform has its own loudness and peak level requirements. Select your target platform β€” the rest is calculated automatically.

Platform LUFS True Peak Sample Rate

Streaming

-14 dB

-1.0 dBTP

unchanged

YouTube

-14 dB

-1.0 dBTP

48 kHz

CD

-12 dB

-0.3 dBTP

44.1 kHz, 16 bit

Streaming β€” universal preset for all music streaming services:

  • Spotify

  • Apple Music

  • YouTube Music

  • Amazon Music

  • Tidal

  • Deezer

  • SoundCloud

Preserves your project’s original sample rate.

YouTube β€” for video content. Converts to 48 kHz, the video platform standard.

CD β€” classic 44.1 kHz / 16 bit format with dithering for quality bit depth reduction.

Tip
 If unsure β€” choose Streaming. This master will work for any platform.

Privacy

The file stays on your device. Nothing is uploaded to the internet. We don’t see your audio, don’t store it, can’t recover it. Close the tab and nothing is left anywhere.

We collect anonymous usage events: which preset was chosen, before/after loudness numbers, processing time, the filename and size you dropped. No IP address, no User-Agent, no cookie, no fingerprint. Used to improve defaults.

About Us

Darwin’s Cat β€” musicians from Berlin. We make rock, and tools for our own work.

Under the Hood β€” Technology & Licenses

This tool stands on the shoulders of several great open-source projects. Big thanks to their authors.

Frequently Asked Questions

What does this tool do?

It picks mastering parameters for your track automatically β€” based on the style, song tempo, desired density and dynamics of the sound β€” for your target platform. The result is a finished WAV master, ready to download.

How does it do mastering?

It takes the audio file, runs a bunch of measurements (loudness, peaks, dynamic range, loudest sections), then automatically picks HPF, EQ, glue-compression and true-peak limiter settings to fit the target platform and style. All locally in your browser.

Does my audio file get uploaded anywhere?

No. Mastering happens entirely in your browser, on your device. The audio bytes never leave your computer. We don't see your file, don't store it, and can't recover it after you close the tab.

Does it replace a human mastering engineer?

For most music that's headed to streaming β€” yes. You get an even level, alignment with platform loudness standards, careful compression and a true-peak limiter. Optional light EQ correction in Advanced.

If you're making an album for a major label, a vinyl release, a soundtrack for Netflix or a single for Eurovision β€” those need artistic decisions an algorithm won't make, and a human mastering engineer is still required. Even there, the tool is useful for intermediate technical mixes.

A bad mix it won't save. A good mix it'll bring to level.

Why call it 'professional' if it’s automatic?

Mastering has two distinct jobs. The first is artistic β€” nuanced tonal decisions, fixing mix problems by ear, taste calls. We don't do that part (apart from optional gentle EQ in Advanced). The second is technical β€” glueing the mix, leveling, true-peak limiting, hitting platform loudness standards, dithering for CD.

The technical part this tool covers with industry-grade algorithms β€” compressors, limiters and equalizers on par with the stock plugins shipped with Cubase, Pro Tools and other DAW leaders. On top of that, our own Darwin's Cat method: automatic tuning of those algorithms to the genre, tempo, dynamics and character of the specific track.

Loudness measurement implements ITU-R BS.1770 / EBU R128 β€” the same standard Spotify, Apple Music, YouTube and broadcasters use to compute LUFS.

Is this an AI tool?

No. We don't use ChatGPT, Claude or any other neural network here. The algorithm is fully deterministic: it measures your track (loudness, dynamic range, loudest sections), looks up the standards for the platform you selected, and computes settings from explicit formulas. Same input + same settings = exactly the same output, every time. All settings are visible and can be manually overridden in Advanced.

How should I prepare the mix before mastering?

Export a WAV so the peaks don't clip on the master bus β€” leave 3+ dB of peak headroom. If the track came out very quiet, that's fine β€” the tool measures the loudest sections and normalizes automatically.

How should I set up the DAW project (sample rate, bit depth)?

Short answer: 48 kHz, 24-bit. Long answer β€” sample rate: 48 kHz is the modern default (streaming, YouTube, video). 44.1 kHz only if the final destination is CD and you'll never stream. 96 kHz only if you have spare CPU and disk β€” marginal real-world benefit in the final mix for most music. Bit depth: 24-bit. Full stop.

Which platform should I choose: Streaming, YouTube or CD?

Streaming is the safe universal choice for Spotify, Apple Music, YouTube Music, Amazon Music, Tidal, Deezer and SoundCloud. Choose YouTube for video β€” the master will be prepared at 48 kHz. Choose CD for the classic 44.1 kHz / 16 bit format with a higher target and dithering.

What LUFS and True Peak targets does mastering use?

Base targets for Streaming / YouTube by Dynamics mode:

  • Dynamic β€” βˆ’14 LUFS (soft, more air)
  • Balanced β€” βˆ’12 LUFS (moderate density)
  • Loud β€” βˆ’10 LUFS (denser, louder)
  • Extreme β€” βˆ’8 LUFS (aggressive, competitive loudness)

For CD, add 2 dB to each number β€” the CD standard is louder than streaming.

True Peak ceiling: βˆ’1.0 dBTP for streaming, βˆ’0.3 dBTP for CD.

Some styles also shift the target by Β±1–2 dB. The final number is computed from platform + style + dynamics and is visible right in the form.

Why does the track not always hit the target LUFS exactly?

The limiter approaches the target iteratively: measure β†’ correct β†’ repeat. It stops either when it hits the target or when going further would only be possible at the cost of clipping and damaged transients. In that case it returns the closest safe version.

What does Style change?

Style picks starting settings for the genre (target LUFS, HPF, attack, release, knee, compressor behavior):

  • Metal, Electronic β€” fast, tight control
  • Jazz, Country β€” softer, more open
  • Pop, Rock β€” universal middle ground

What do Dynamics and Density change?

Dynamics controls overall density and loudness of the track:

  • Dynamic β€” more air, natural dynamics
  • Balanced β€” moderate density, the golden middle
  • Loud β€” denser and louder, for streaming-oriented genres
  • Extreme β€” maximum density, competitive loudness

Density controls the compressor β€” how firmly it holds the mix together:

  • Light β€” gentle gluing
  • Medium β€” moderate cohesion
  • Heavy β€” dense leveling

You can combine them: Light + Dynamic β€” natural; Heavy + Extreme β€” maximum density.

When should I use Advanced settings?

Use them when the automatic result is close but you want to fine-tune it: a little less compression, different attack/release, another true peak ceiling, more or less Air, or less Mud. If you are not sure, keep the automatic values β€” they already account for style, dynamics and the track analysis.

Why do I need a true-peak limiter and oversampling?

True Peak protects against inter-sample peaks that can appear after AAC/MP3/Opus encoding or D/A conversion. 4Γ— or 8Γ— oversampling helps the limiter see those peaks in advance and keep them below the ceiling.

Why can the mastered track sound different, not just louder?

Mastering changes more than level. HPF removes unnecessary low end, EQ adjusts tonal balance, the compressor glues the mix, and the limiter controls peaks. That changes density, attack and the feel of bass and treble. Compare the original and the master at similar loudness β€” otherwise the louder version almost always feels better.

What should I do if the master sounds too squashed?

Try a softer Dynamics mode, reduce Density, raise the threshold offset or lower the target LUFS. For live-sounding genres it's often better to keep dynamics than chase maximum loudness β€” streaming platforms will normalize overly loud tracks down anyway. Better to give up a couple of dB of loudness β€” you gain in character.

Can I make several versions of one track?

Yes. Change platform, style, Dynamics, Density or Advanced parameters and run mastering again without re-loading the file. Each result is saved on the page, so you can compare versions and download the one that sounds best.

Which browsers are supported?

Any modern desktop browser β€” Chrome, Firefox, Safari, Edge. Mobile browsers technically work but processing is much slower there; a desktop is recommended for tracks longer than a minute or two.

What’s the file size limit?

About 200 MB or 15 minutes of audio. Browser memory is limited; longer or heavier files may fail to allocate. For shorter clips there's no practical limit beyond your CPU patience.

Do you collect any data?

Only anonymous usage events: which preset was chosen, before/after loudness numbers, processing time, the filename and size you dropped. No IP address, no User-Agent, no cookie, no fingerprint. Used to improve defaults. The audio bytes themselves are never transmitted anywhere.

Comments

Be the first to leave a comment.

By submitting you agree to our privacy notice.