REFERENCE LIBRARY / EVIDENCE & LISTENING
Behind the
experiments.
The listening pages let you hear a difference. This is where we explain what changed, what we measured, and what we can reasonably conclude. The goal isn't to crown a winner. It's to make the cause of an audible change easier to understand.
01 / MIXING
One performance, three deliberately different spaces
The Mixing Lab starts with the same 30 seconds of music (0:40–1:10) from 14 separate recorded parts, called stems. These are processed-dry parts, not untouched microphone recordings. We made three mixes and adjusted each finished example to the same measured average loudness, −20 LUFS. That makes it easier to compare the creative differences without being swayed by one mix simply playing louder. We did not apply mastering compression or limiting to these versions.
| Mix | Side / mid | What the treatment illustrates |
|---|---|---|
| Extreme narrow | −42.77 dB | An intentionally near-mono presentation with much of the sound gathered near the center |
| Conventional reference | −14.61 dB | A more natural stereo distribution using the same recorded performance |
| Creative / polished | −9.68 dB | Deliberately theatrical width, motion, ambience and arrangement emphasis |
These numbers tell us how much more sound energy is sent toward the sides than toward the centre of the stereo picture. They cannot tell us how far back an instrument seems, whether individual parts are easier to follow, or which mix sounds best. We changed several things at once, including left-right placement (panning), echo/room effects and automated moves. So this is a listening demonstration, not a controlled test of just one setting.
Why playback can change the sense of space
Stereo width and front-to-back depth are not the same thing. A mix may have clearly identifiable left/right placement while its apparent depth changes with the listening setup. Headphones send a separate signal to each ear, while loudspeakers interact with the room. Placement, reflections and listening position can influence perceived depth, centre imaging and how spatial movement is heard.
For this demonstration, headphones may make some stereo cues easier to follow, but neither playback method is a universal reference for every spatial effect. Comparing headphones and speakers at comfortable levels can illustrate an important principle: the recorded mix is only one part of the listening experience. This is a general explanation, not a conclusion from a controlled headphone-versus-speaker study.
Try the three-mix comparison → Read the mixing and soundstage lesson →
02 / MASTERING
Same rough mix. Different finishing decisions.
Our mastering comparison is built around one shared rough mix. That matters: if you change the mix between examples, listeners cannot know whether a missing triangle, a buried guitar or a changed vocal balance was caused by mastering or by a different mix.
Keep the movement
Preserve greater contrast between quiet details and musical peaks, rather than chasing maximum average loudness.
Some control, some punch
Use a more restrained combination of level, equalization and dynamic control to show a middle ground.
Push it far enough to hear
Extreme processing reduces the contrast between sharp musical hits and the sound around them. That lets you hear the trade-off when making the whole song consistently loud becomes the priority.
These are teaching extremes, not claims that every loud master is bad or every dynamic version sounds best. A bright percussion accent, such as the triangle in the rough mix, is useful to listen for. Does its attack stand out, get masked, or seem to change its place among the other instruments?
How to judge the comparison fairly
Listen two ways. At their actual mastered levels, you hear the practical loudness difference. At approximately matched listening loudness, you can pay closer attention to punch, distortion, tonal balance and whether quieter details are easier to follow. Louder often seems more exciting by itself, so the two listening modes answer different questions.
Look at integrated loudness (average level over the passage), peak and true-peak level (how close the strongest moments get to the ceiling), and measures of peak-to-average contrast. No single meter gives a universal score for musical quality. If EQ changes as well as compression, don't attribute every tonal difference to compression alone.
Try the Mastering Lab → Explore the mastering and dynamic-range lesson → Read the reference guide →
03 / THE METHOD
What makes a listening comparison useful?
Hold the right things constant
Use the same performance, time range and starting file when testing a specific production decision. Label any exceptions rather than hiding them.
Distinguish gain matching from mastering
Our three mixing examples were gain-matched for listening; that operation did not make them separately mastered records. For mastering, report the chosen gain and dynamic treatment of each version.
Let your ears and the measurements ask different questions
A stereo ratio tells us about side energy. A loudness meter tells us about level. Listening impressions can describe spaciousness or punch. They support one another, but none replaces the others.
Report limitations
These are intentionally exaggerated educational mixes and masters. Informal listening on a particular system is not a blinded, representative study, and an exaggerated demonstration is not proof that one approach suits every song.
04 / CREDITS & SOURCES
Where the performance came from
The mixing excerpt is from “Let The Light In” by Tom McKenzie, from Extraordinary (2025). The stems came from AlbumDB: A multitrack dataset of a ten-song album with stereo and immersive 7.1.4 masters, credited to Thomas McKenzie, Eloi Moliner Juanpere and Alec Wright. The dataset lists a CC BY 4.0 licence; see also the AlbumDB research paper (AES Paper 10278). Our educational edits are independent; no endorsement is implied.
To understand the surrounding concepts, continue with the listening-language glossary, mixing and soundstage, and mastering and dynamic range. The interactive demonstrations belong in the Lab; the reasoning belongs here.