top of page

Equalization  (July 2017)

 

Perhaps you can recall the days when radios and record players had a “Tone” control.  It allowed the listener to place greater emphasis on higher frequencies (treble) or lower frequencies (bass).  Decades ago, it gave way to separate “Treble” and “Bass” controls that permit one to independently emphasize or de-emphasize high and low frequencies, respectively, and those controls remain a mainstay on modern audio equipment.  Adjusting the Tone, Treble, or Bass controls, or more generally changing the amplitude of different audio frequency bands, is known as “equalization.”

 

For those who want greater precision in adjusting tone than the Treble and Bass controls provide, a graphic equalizer can be employed (Figure 1).  It can either be a separate component, or it may be integrated within the amplifier.  Instead of just the Treble and Bass controls, graphic equalizers have several—four to ten for consumer-grade equipment, 25 or 31 for professional-grade equalizers—where each knob or slider adjusts the amplitude of a specific frequency band.  The greater the number of controls, the narrower the frequency band, and the greater precision one has in getting just the “right sound.”

Figure 1.  Software 10-band graphic equalizer.

 

Another type of equalizer is known as a parametric equalizer (Figure 2).  Whereas a graphic equalizer’s bands are set at predetermined frequency ranges, parametric equalizers allow one to specify the center frequency and the width of each band independently.  This enables even greater precision in eliciting the sound one is after.  Music-playing software that includes equalization capability emulates either graphic or parametric equalizers, depending on the program.

Figure 2.  Software parametric equalizer.

 

So, what does equalization have to do with conversion and noise reduction? 

 

Conversion

For conversion, the answer most of the time is “nothing.”  Equalization is unnecessary for records manufactured later than the mid-1950s and for tapes.  The sound engineer adjusted equalization during the production of the original recording, and the conversion process should preserve the tonal characteristics of the original recording.  (When playing music, the listener of course can adjust tone or equalizer controls in accordance with the listening environment and individual preferences.)

 

Equalization is relevant, however, in converting records produced before the mid-1950s—78s, LPs, and 45s.  The story is a complex one, and it requires a brief, and for the moment oversimplified, digression.  For reasons that will be explained in the following paragraphs, record companies have applied equalization since the mid-1950s according to what is known as the “RIAA curve,” or “RIAA pre-emphasis” (Recording Industry Association of America):  treble is boosted, and bass is attenuated.  Sound systems from stand-alone record players to component systems with separate turntables, preamps, and amplifiers apply an inverse equalization (termed “RIAA de-emphasis”) that reverses the RIAA curve so what the listener hears matches the tonal balance of the original music.  End of digression and back to pre-1950s records.

 

Recording technology prior to 1926 was mechanical in nature.  Sound was captured directly onto the recording medium via a horn, diaphragm, and needle.  It was capable of recording frequencies from 168 Hz (roughly F below middle C) to 2 KHz (three octaves above middle C), and there was neither the need nor the technology to equalize the sound.  Consider what happens when one plays a pre-1926 78 on a modern system.  Since the system was designed to play records where RIAA pre-emphasis has been used, it will apply RIAA de-emphasis to remove the pre-emphasis.  But since no pre-emphasis had been applied during the record’s manufacture, one ends up with a sound that is bass-heavy and treble-light.  Consequently, when I convert a pre-1926 78, I reapply the RIAA pre-emphasis, thereby undoing what my sound system has done, and the record will sound like it was supposed to.

 

With the change from mechanical to electrical recording in 1925, especially the introduction of microphones, the frequency range expanded downward from 168 Hz to 60 Hz (B two octaves below middle C) and upward from 2 KHz to 6 KHz (roughly F# above the highest C on a piano).  A problem immediately arose:  The expanded low frequency range produced large vibrations in cutting heads that could damage groove walls and bleed low frequencies into adjacent grooves.  Moreover, they could affect the tonearm during playback, sometimes causing it to skip to an adjacent groove.

 

Enter equalization.  In 1926, Western Electric patented a system that “rolled off” bass frequencies below the “turnover frequency” of 250 Hz. (“Rolled off” means increasing attenuation of the bass as the frequency decreases—frequencies just below 250 Hz are attenuated just a bit while those at the low end of the frequency range are attenuated a lot; the “turnover frequency” is the frequency below which rolloff is applied).  This system was the only one in existence and thus the de facto standard at the time, and Western Electric received royalty payments from any record company that used it.

 

In 1929, a different equalization system was developed that avoided infringing on the Western Electric patents, as well as paying the accompanying royalties.  Unlike Western Electric’s system, however, the turnover frequency wasn’t fixed.  This allowed record companies to select their own turnover frequencies according to what they thought was the best frequency for the average record player, and that’s what they did.  Without standards or conventions in place, here’s what occurred.

  • Turnover frequencies ranged widely—between 250 Hz and 1 KHz.

  • Many recording companies had a “pet” turnover frequency, others had more than one, and still others didn’t apply bass rolloff at all.

  • Many record companies regarded their turnover frequencies as proprietary trade secrets and never informed the public of what they were.

  • While the major record companies tended to use their own recording studios and could dictate or at least recommend the turnover frequency to be employed, smaller companies used what independent studios produced, and the studios—or sometimes individual sound engineers—determined which turnover frequency to apply.

  • Some labels released records using other companies’ masters, or sometimes purchased records and released them after replacing the original labels with their own.  Information about the turnover frequencies was lost.

 

The upshot of all this was that consumers rarely had any knowledge of turnover frequencies or what to do about them, and given the wide variety in use, a 78 could sound okay, bass-heavy, or bass-light.  As electric record players overtook mechanical gramophones for home use, the introduction of the Tone control provided the ability to accommodate different turnover frequencies.  (And now you know the genesis of the Tone control.)

 

But there’s more.  In addition to excessive cutting head and tonearm vibration resulting from the extended low frequency range, treble frequencies were overshadowed by bass frequencies and, on playback, were comparatively weak.  About 1937, record companies discovered that they could solve this problem, and reduce surface noise at the same time, by boosting treble, usually by +3 or +6 dB/octave.  (This is the origin of pre-emphasis, but since “RIAA pre-emphasis” commonly refers to both increasing treble strength and decreasing bass strength, I’ll use “treble pre-emphasis” to refer to the former.)  There still weren’t any standards or conventions, and treble pre-emphasis—anywhere from +0 dB/octave (no pre-emphasis) to a whopping +16 dB/octave—led to the same tonal variability and consumer frustration that bass rolloff had.

 

The situation became more complex with the introduction of magnetic tape into the recording process after World War II, as it expanded the frequency range to 10 Hz - 20 KHz, and with the 1948 introduction of LPs, where microgrooves had even less tolerance for large bass vibrations than 78s.  By necessity, the production of LPs employed bass rolloff and treble pre-emphasis, and in the U.S. the production of 78s followed suit.  The introduction of 45s in 1949 muddied things further, as they employed their own bass rolloff and treble pre-emphasis parameters.

 

Consumers clamored for standardization, and record companies realized that a single convention was to their advantage as well.  In 1952, RCA introduced an equalization curve for recording that combined treble pre-emphasis (beginning at 2.1 KHz [two octaves above middle C], +13.7 dB measured at 10 KHz) and a bass turnover frequency of 500 Hz.  There was a concomitant de-emphasis curve (Figure 3) to remove the pre-emphasis (with the addition of a Low Shelf filter at 50 Hz to avoid amplifying low-frequency noise such as turntable rumble).  RCA’s equalization curve was adopted by RIAA in 1953, and most U.S. companies gravitated toward it between the mid-1950s and the early 1960s.

Figure 3.  RIAA de-emphasis equalization curve

(+dB gain/-dB attenuation vs. frequency).

 

To summarize:

  • Pre-1926 78s:  no equalization.

  • 1926-1929 78s:  Western Electric bass attenuation turnover frequency—250 Hz (for those companies who used it).

  • 1929-1937 78s:  Numerous turnover frequencies used (if any was used at all).

  • 1937-1948 78s:  Numerous turnover frequencies and numerous treble pre-emphasis curves used.

  • 1948-ca. 1955:  Numerous turnover frequencies and numerous treble pre-emphasis curves used for both 78s and LPs.

  • 1949-ca.1955:  45s introduced their own turnover frequency and treble pre-emphasis curve.

  • 1953:  RCA’s bass rolloff/treble pre-emphasis curve adopted by RIAA.

  • ca. 1955-early 1960s:  U.S. 78s, 45s, and LPs increasingly used the RIAA curve to the point where it became the de facto standard.  European companies followed later (well into the 1960s), and some Russian and Asian companies didn’t adopt the RIAA curve until 1975.

 

Given this background, there is a dizzying array of possibilities when converting records manufactured before the mid-1950s.  Fortunately, several groups have researched historical records and published tables of bass rolloff and treble pre-emphasis parameters used by many recording companies, and the years in which they were used.  When they’re available on-line, I can download and install a de-emphasis curve appropriate for the record I’m working on; otherwise, I can use the tabled parameters to create and apply my own de-emphasis equalization.

 

If there is nothing published that’s relevant to my current project, or if there is disagreement among published tables, I can try different turnover frequencies (starting at 250 or 300 Hz and increasing it as required) and different treble pre-emphasis parameters (starting at 2.1 KHz with -6 or -8 dB attenuation at 10KHz and experimenting with different attenuation values).  In the end, the “keeper” is what sounds the best.

 

Noise Reduction

While I normally forego equalization in noise reduction, there are certain situations where it can be a useful tool.  With pre-1945 records, for example, audio signals occurring outside the frequency range that the recording technology of the time could capture are, by definition, noise, and such often appear with worn records.  Applying equalization to attenuate frequencies above and below the technology’s upper and lower limits, respectively, will significantly remove noise that doesn’t belong in the recording.

 

Another situation where I have used equalization in noise reduction is with voice recordings made with a portable microcassette recorder.  In addition to capturing conversations, these recorders also capture their own transport noise, which usually manifests itself as a high-pitched whine.  I address this by severely decreasing the amplitude in a single, very narrow frequency band.  One of the algorithms I use includes an adjustment of the equalization’s slope (dB/octave) in addition to frequency, bandwidth, and amplitude, and this permits finer tuning of the frequency band to be altered by the equalization.  This usually works well:  It removes the whine, and the voices still sound natural.

 

With worn records, I will experiment with equalization if I hear something that I don’t believe belongs there or if I spot what might be an anomaly in a spectral analysis display.  I apply the technique I just described, but very cautiously.  In some cases, the music benefits from removal of pervasive noise in a narrow frequency band, but in other cases this approach has yielded undesirable changes in the music.

 

As was the case with the noise reduction algorithms discussed in a previous posting (What Happens in Noise Reduction for a Record), I can turn equalization on and off while previewing the audio track.  This provides indications of how successful the equalization will be in removing the noise while preserving the tonal character of the music.

Back to Paul's Blog & Contents

bottom of page