Master Your Audio with Informed Scientific Decisions

Mastering music is incredibly interesting and, on occasion, equally challenging. It is a technical job designed to translate a recording across multiple mediums and playback systems.

At the same time, it is a creative process that often requires an emotional understanding of an artist’s intentions. And in many ways, it is also a scientific practice, one that benefits from in-depth knowledge of physics, biology, and even anthropology from time to time.

Beyond improving masters through trial and error, reviewing research papers and controlled studies can reveal patterns worth considering during real-world sessions. These findings range from psychoacoustic research to blind listener preference tests and even historical data drawn from commercially successful releases.

What follows is a structured examination of audio mastering science, using measured research to inform practical decisions—without replacing critical listening or artistic intent.

Measured Standards for Dynamic Range

One of the most concrete areas where research intersects with mastering is dynamic range.

A 2014 paper by Pedro Pastana and Joshua D. Rice, presented at the 53rd AES Convention, analyzed 928 commercially successful songs to identify measurable patterns in dynamics. Rather than focusing on loudness alone, the study examined peak-to-RMS ratios across frequency ranges.

Understanding Peak-to-RMS Ratio

  • Peak represents the highest amplitude of a signal.
  • RMS (Root Mean Square) represents the average loudness over time.
  • Comparing the two reveals the crest factor, or the amplitude difference between peaks and perceived loudness.

The higher the ratio, the greater the distance between peak energy and average loudness.

This relationship provides insight into how controlled—or uncontrolled—the dynamics are within specific frequency ranges.

Dynamic Range by Octave

The study measured average peak-to-RMS ratios across octaves and revealed a clear trend:

  • Low frequencies exhibit the most controlled dynamics
    • Peak-to-RMS ratio: ~16
  • High frequencies exhibit the least controlled dynamics
    • Peak-to-RMS ratio: ~28

Visually, this means low frequencies tend to have smaller crest factors, while high frequencies display larger crest factors, reflecting sharper transient behavior and less compression.

Grouping Octaves for Practical Mastering

Rather than working with eight individual octaves, it often makes more sense to consolidate them into three broader frequency ranges:

  1. Low frequencies
  2. Mid frequencies
  3. High frequencies

This approach allows any processing changes to occur over wider frequency spans, making them less audible and more natural sounding.

Measuring and Adjusting Crest Factor

A practical workflow may involve:

  • Splitting the signal into three frequency ranges
  • Soloing each range
  • Measuring peak and RMS values
  • Comparing the resulting ratio to research benchmarks

From there:

  • If the ratio is too high, compression may be appropriate
  • If the ratio is too low, expansion may restore life
  • Afterward, the overall response is adjusted to maintain the original intent

This process is time-consuming and not something to perform on every session. However, attempting it even once can be extremely valuable.

It establishes a reference for what is considered “somewhat standard” in mastering—giving you the option to either adhere to it or intentionally deviate.

Expected Stereo Width and Channel Balance

Stereo width is another area where research confirms long-standing best practices.

Monocompatibility as a Standard

For over four decades, monocompatibility has remained widely accepted. Early stereo recordings experimented with extreme and unconventional panning, but those practices largely disappeared.

Research measuring left and right channel RMS values found:

  • Maximum deviation between channels: 0.8 dB

This does not mean a master cannot feel wide. It means that width is typically achieved without imbalanced channel energy.

What This Implies for Mastering

Several practical conclusions emerge:

  • Powerful and essential elements should remain centered
    • Lead vocal
    • Bass
    • Kick
    • Snare
  • Avoid processing that affects only one channel
  • Keep limiting linked or mostly linked to prevent uneven attenuation

Maintaining similar RMS values in both channels preserves focus, impact, and translation.

Stereo Expansion: What Works and What Doesn’t

Stereo expansion is not inherently problematic—but how it is achieved matters.

Recommended Practices

  • Stereo expansion above 200 Hz is generally acceptable
  • Changes should be subtle
  • Focus on adjusting the side image, not creating artificial differences

By amplifying the side image, mastering enhances differences already present in the mix rather than introducing new ones that could destabilize the center image.

What to Avoid

  • Delay-based stereo expanders
  • Processing that introduces phase interference
  • Techniques that widen the image by displacing centered elements

Artificial widening often causes important centered instruments to lose focus.

Preferred Reverb Levels in Listener Studies

Reverb is often thought of as a mixing concern, but mastering can significantly affect how reverb is perceived.

Listener preference studies show that:

  • Dry signals are generally preferred
  • The most favored reverb level is 9 loudness units below the dry signal

This balance offers spatial depth without overwhelming clarity.

Why Reverb Becomes More Noticeable During Mastering

Reverb consists largely of quiet reflections that are often masked by louder sounds. During mastering, processes such as:

  • Limiting
  • Clipping
  • Compression with makeup gain
  • Ceiling-based loudness increases

all amplify these quieter elements.

As a result:

  • Reverb that felt subtle in the mix can become overly prominent
  • Dynamic control often unintentionally increases perceived ambience

The Dynamic Range–Reverb Relationship

This issue becomes especially apparent when controlling peak-to-RMS ratios.

Consider the following scenario:

  1. Mid frequencies contain noticeable reverb
  2. The peak-to-RMS ratio is higher than desired
  3. Compression is applied to control peaks
  4. Makeup gain restores overall level
  5. Quiet details—including reverb—are amplified

The result is a more obvious reverb tail, even though no reverb was added.

Unfortunately, there is no easy way to measure this relationship precisely when working with a stereo file.

Best Practical Solution

The most effective solution occurs before mastering:

  • Mix engineers slightly lower reverb and temporal effects
  • This accounts for the inevitable increase in perceived ambience during mastering

While mastering engineers can manage dynamics, they cannot easily separate dry signal from reverb in a stereo track.

Preparing for mastering during the mix stage leads to better translation and fewer compromises.

By integrating research-backed insights into dynamic range, stereo width, and reverb perception, mastering decisions can be made with greater confidence—balancing measurable standards with critical listening and artistic intent.