Can EQ curve differences be measured objectively?

Last updated: June 14, 2026 Reading time: approx. 8 min

What recording EQ curve was used on a given LP, and how does a reissue or remaster differ from the original? When you try to answer questions like these with objective measurement rather than subjective listening, LTAS (Long-Term Average Spectrum) comes up as a candidate tool. The Liner Notes II series on this site (Liner Notes II-a / Liner Notes II-b / Liner Notes II-c) took 1950s Capitol records as the material and showed, through concrete examples, what this tool can and cannot see. This FAQ distills those points into a single overview.

Short answer

Not completely. But if the tool's scope is constrained and the conditions are aligned, some limited answers can be obtained. What LTAS shows relatively strongly is:

The differences in production choices among releases derived from the same session master, expressed numerically band by band
Indirect corroboration of the cutting curve (whether AES or RIAA, for example)

That is about as far as LTAS can go. Beyond it, four caveats always follow: physical variation on the cutting side, individual differences on the pressing side, the influence of the playback environment, and the inherent limits of LTAS itself.

The tool: what LTAS is

LTAS (Long-Term Average Spectrum) is the energy distribution of a signal averaged across frequency over a given time span. It expresses the spectral balance of the sound as a single curve, in the form of "across this entire song, how strong is the 1 kHz component, how about 5 kHz, how about 10 kHz".

When you take multiple releases derived from the same session master (the original pressing, reissues, remasters, digital transfers) and put them through LTAS under the same conditions, the band-by-band differences become visible numerically. Liner Notes II-b lined up five releases derived from H-488 "Songs For Young Lovers" and read the mastering direction numerically: the 12-inch reissue W-587 sits +6 to +9 dB brighter than the 10-inch initial pressing in the 5–12 kHz range, and the 1998 digital remaster has roughly +4 dB more low-end energy.

However, LTAS is an "average," so movements of sound that happen within short time windows (the attack of an instrument, for instance) are smoothed out and become hard to see. This will be addressed concretely later under "What LTAS itself cannot show".

Four caveats in the measurement

Taking the differences visible in LTAS and directly tying them to "what cutting curve a given pressing used," or asserting with finality that "the remaster's production intent was X," requires caveats on each of the following four axes.

Cutting-side variation

Cutting the same session master multiple times through the same cutting chain does not produce identical results. Cutter-head temperature, stylus wear, amplifier stability, the room temperature and humidity of the day, the engineer's small adjustments, and other physical factors each appear in the result as roughly ±1–2 dB of variation.

The comparison of the initial pressings D2 and D4 of H-488, covered in Liner Notes II-a, was precisely a case of "variation between different cuttings, presumed to be under the same curve, made around the same time". When the LTAS-visible difference falls within ±1–2 dB, telling whether it is physical variation in the cutting equipment or a substantive difference upstream in the master cannot be done from a single observation.

Candidates for the cutting curve can be tested by switching AES and RIAA on the playback side or by applying digital conversion and checking whether the theoretical residual shrinks. But finer details — individual-unit characteristics of the cutter head, the adjustments of the day — only become reachable with the help of primary sources such as published equipment specifications or session logs.

Pressing-side variation

Thousands of pressings come off a single stamper, but stampers wear with use. Between a pressing made from a brand-new stamper and one made just before stamper replacement, the sharpness of the groove edges differs subtly, and the high-frequency response may shift by a few dB.

In addition, in the early 1950s the quality and uniformity of PVC ("biscuits") was not yet industrially mature; surface-noise quality and material stability are believed to have varied considerably from biscuit to biscuit, and from plant to plant. Such plant-side quality differences cannot be ruled out as a factor influencing LTAS results, even between pressings derived from the same session master.

Quantitatively separating pressing variation and material variation from other contributions in LTAS would require transcribing many pressings from the same matrix under the same conditions and comparing them. From a single observation, one cannot go beyond "the possibility cannot be ruled out".

Playback-side influence

Listening to an LP always means going through some combination of cartridge, tone arm, and a specific pressing. And the characteristics of this playback side appear directly in the LTAS result. The cartridge's high-frequency resonance, stylus geometry (spherical, microline, etc.), tone-arm inertia, and the interaction between groove shape and stylus contact point combine such that the high-frequency response retrieved from an LP can move on the order of several dB.

Even on the same pressing through the same equipment, systematic differences can emerge between transcriptions done on different days. In the example shown in Liner Notes II-c, where the same side of H-477 was transcribed twice on different days, a systematic difference of +1.4 dB at 1 kHz and +1.8 dB on average across the presence band (4–8 kHz) emerged even after subtracting the playback-EQ difference. Tracking-force adjustment or the cleaning state of the pressing that day are plausible candidates, but the cause has not been pinned down.

For claims at the 1 dB level, the comparison transcriptions need to be made consecutively on the same day, ideally with downstream digital variants derived from a single transcription pass.

What LTAS itself cannot show

LTAS is a tool for viewing "energy averaged across frequency over a given time window". Averaging loses certain information. Specifically, short-duration attacks such as a brush stroke or a stick strike are smoothed out and become hard to see clearly.

The example of B Sharp Blues, covered in Liner Notes II-c, illustrates this well. Comparing T-477 with the master-tape digital transfer (Mosaic Records, the closest available external reference to the master) in LTAS shows only a modest difference of +2.08 dB in the 4–8 kHz mean on B Sharp Blues. But focusing on the instant of a stick attack on the drums and computing the sharpness of the attack (the crest factor, the ratio of the peak value to the average value of a signal) reveals a clear difference of +3.96 dB. For the same two pressings, LTAS says "about the same," and the attack analysis says "noticeably different". Neither is wrong; they are simply looking at different facets.

A more rigorous verification design would call for transcribing many pressings from the same matrix under the same conditions and comparing them. In practice, however, gathering many pressings in matched condition and running them through a single environment is difficult, and this series has proceeded by combining single-observation analyses.

What can still be said

"LTAS is a tool for indicating tendencies, not for settling questions." Writing it that way may sound like a compromise, but it is the reverse. Knowing the limits of the tool is precisely what makes it possible to draw the line between what can be stated outright and what should be written as conjecture.

What LTAS cannot capture can be supplemented piece by piece with other tools — observing the leader-tape traces of the cutting master in spectrograms (which show time and frequency simultaneously) to work backward through the lineage, reading dynamic emphasis from attack envelopes, gathering many pressings from the same matrix for comparison. The three-part Liner Notes II series (II-a / II-b / II-c) was written as a first step in combining such tools.

To the question "Can it be measured objectively?", the author's answer is: "Within a certain range, with the conditions aligned, yes — and the range of what can be measured should itself be written down honestly".

Revision History

June 14, 2026: Initial publication