The Audio Expert: Audio Fidelity, Measurements, And Myths—Part 1
Only four parameters are needed to define everything that affects the fidelity of audio equipment

May 02, 2012, by Ethan Winer


Here we present a portion of a chapter in the new book “The Audio Expert” by Ethan Winer, published by Focal Press.


“Science is not a democracy that can be voted on with the popular opinion.” — Earl R. Geddes, audio researcher

In this chapter I explain how to assess the fidelity of audio devices and address what can and cannot be measured. Obviously, there’s no metric for personal preference, such as intentional coloration from equalization choices or the amount of artificial reverb added to recordings as an effect. Nor can we measure the quality of a musical composition or performance.

While it’s easy to tell—by ear or with a frequency meter—if a singer is out of tune, we can’t simply proclaim such a performance to be bad. Musicians sometimes slide into notes from a higher or lower pitch, and some musical styles intentionally take liberties with intonation for artistic effect.

So while you may not be able to “measure” Beethoven’s Symphony #5 to learn why many people enjoy hearing it performed, you can absolutely measure and assess the fidelity of audio equipment used to play a recording of that symphony. The science of audio and the art of music are not in opposition, nor are they mutually exclusive.

High Fidelity Defined

By definition, “high fidelity” means the faithfulness of a copy to its source. However, some types of audio degradation can sound pleasing—hence the popularity of analog tape recorders, gear containing tubes and transformers, and vinyl records. As with assessing the quality of music or a performance, a preference for intentional audio degradation cannot be quantified in absolute terms, so I won’t even try. All I can do is explain and demonstrate the coloration added by various types of audio gear and let you decide if you like the effect or not.

Indeed, the same coloration that’s pleasing to many people for one type of music may be deemed unacceptable for others. For example, the production goal for most classical (and jazz or big band) music is to capture and reproduce the original performance as cleanly and accurately as possible. But many types of rock and pop music benefit from intentional distortion ranging from subtle to extreme.

“The Allnic Audio’s bottom end was deep, but its definition and rhythmic snap were a bit looser than the others. However, the bass sustain, where the instrumental textures reside, was very, very good. The Parasound seemed to have a ‘crispy’ lift in the top octaves. The Ypsilon’s sound was even more transparent, silky, and airy, with a decay that seemed to intoxicatingly hang in the air before effervescing and fading out.” — Michael Fremer, comparing phonograph preamplifiers in the March 2011 issue of Stereophile magazine

Perusing the popular hi-fi press, you might conclude that the above review excerpt presents a reasonable way to assess and describe the quality of audio equipment. It is not. Such flowery prose might be fun to read, but it’s totally meaningless because none of those adjectives can be defined in a way that means the same thing to everyone. What is rhythmic snap? What is a “crispy” lift? And how does sound hang in the air and effervesce?

In truth, only four parameters are needed to define everything that affects the fidelity of audio equipment: noise, frequency response, distortion, and time-based errors. Note that these are really parameter categories that each contain several subsets. Let’s look at these categories in turn.

The Four Parameters

Noise is the background hiss you hear when you raise the volume on a hi-fi receiver or microphone preamp. You can usually hear it clearly during quiet passages when playing cassette tapes. A close relative is dynamic range, which defines the span in decibels (dB) between the residual background hiss and the loudest level available short of gross distortion.

CDs and DVDs have a very large dynamic range, so if you hear noise while playing a CD, it’s from the original master analog tape, it was added as a by-product during production, or it was present in the room and picked up by the microphones when the recording was made.

Subsets of noise are AC power-related hum and buzz, vinyl record clicks and pops, between-station radio noises, electronic crackling, tape modulation noise, left-right channel bleed-through (cross-talk), doors and windows that rattle and buzz when playing music loudly, and the triboelectric cable effect. Tape modulation noise is specific to analog tape recorders, so you’re unlikely to hear it outside of a recording studio.

Modulation noise comes and goes with the music, so it is usually drowned out by the music itself. You can often hear it on recordings that are not bright sounding, such as a bass solo, as each note is accompanied by a “pfft” sound that disappears between the notes. The triboelectric effect is sometimes called “handling noise” because it happens when handling poor-quality cables. The sound is similar to the rumble you get when handling a microphone. This defect is rare today, thanks to the higher-quality insulation materials used by wire manufacturers.

Frequency response describes how uniformly an audio device responds to various frequencies. Errors are heard as too much or too little bass, midrange, or treble. For most people, the audible range extends from about 20 Hz at the low end to slightly less than 20 KHz at the high end. Some youngsters can hear higher than 20 KHz, though many senior citizens cannot hear much past 12 KHz.

Some audiophiles believe it’s important for audio equipment to pass frequencies far beyond 20 KHz, but in truth there’s no need to reproduce ultrasonic content because nobody will hear it or be affected by it. Subsets of frequency response are physical microphonics (mechanical resonance), electronic ringing and oscillation, and acoustic resonance. Resonance and ringing will be covered in more detail later in this and other chapters.

Distortion is a layman’s word for the more technical term nonlinearity, and it adds new frequency components that were not present in the original source. In an audio device, non-linearity occurs when a circuit amplifies some voltages more or less than others, as shown in Figure 2.1. This nonlinearity can result in a flattening of waveform peaks, as at the left, or a level shift near the point where signal voltages pass from plus to minus through zero, as at the right. Wave peak compression occurs when electrical circuits and loudspeaker drivers are pushed to levels near their maximum limits.

Figure 2.1: Two types of nonlinearity: peak compression at the top and/or bottom of a wave (left), and crossover distortion that affects electrical signals as they pass through zero volts (right). (click to enlarge)

Some circuits compress the tops and bottoms equally, which yields mainly odd-numbered harmonics—3rd, 5th, 7th, and so forth—while other circuit types flatten the top more than the bottom, or vice versa. Distortion that’s not symmetrical creates both odd and even harmonics—2nd, 3rd, 4th, 5th, 6th, and so on. Crossover distortion (shown in Figure 2.1) is also common, and it’s specific to certain power amplifier designs. Note that some people consider any change to an audio signal as a type of distortion, including frequency response errors and phase shift. My own preference is to reserve the term “distortion” only when nonlinearity creates new frequencies not present in the original.

When music passes through a device that adds distortion, new frequencies are created that may or may not be pleasing to hear. The design goal for most audio equipment is that all distortion be so low in level that it can’t be heard. However, some recording engineers and audiophiles like the sound of certain types of distortion, such as that added by vinyl records, transformers, or tube-based electronics, and there’s nothing wrong with that. My own preference is for gear to be audibly transparent, and I’ll explain my reasons shortly.

The two basic types of distortion are harmonic and intermodulation, and both are almost always present together.

Harmonic distortion adds new frequencies that are musically related to the source. Ignoring its own inherent overtones, if an electric bass plays an A note whose fundamental frequency is 110 Hz, harmonic distortion will add new frequencies at 220 Hz, 330 Hz, 440 Hz, and subsequent multiples of 110 Hz. Some audio devices add more even harmonics than odd, or vice versa, but the basic concept is the same.

In layman’s terms, harmonic distortion adds a thick or buzzy quality to music, depending on which specific frequencies are added. The notes created by most musical instruments include harmonics, so a device whose distortion adds more harmonics merely changes the instrument’s character by some amount.

Electric guitar players use harmonic distortion—often lots of it—to turn a guitar’s inherent plink-plink sound into a singing tone that has a lot of power and sustains.

Intermodulation distortion (IMD) requires two or more frequencies to be present, and it’s far more damaging audibly than harmonic distortion because it creates new sum and difference frequencies that aren’t always related musically to the original frequencies.

For example, if you play a two-note A major chord containing an A at 440 Hz and a C# at 277 Hz through a device that adds IM distortion, new frequencies are created at the sum and difference frequencies:

Sum: 440 Hz + 277 Hz = 717 Hz
Difference: 440 Hz + 277 Hz = 163 Hz

717 Hz is about halfway between an F and F# note, and 163 Hz is slightly below an E note. Neither of these are related musically to A or C#, nor are they even standard note pitches. Therefore, even in relatively small amounts, intermodulation distortion adds a dissonant quality that can be unpleasant to hear. Again, both harmonic and intermodulation distortion are caused by the same nonlinearity and thus are almost always present together. What’s more, when IM distortion is added to notes that already contain harmonics, which is typical for all musical instruments, sum and difference frequencies related to all of the harmonics are created, as well as for the fundamental frequencies.

Another type of distortion is called aliasing, and it’s unique to digital audio. Like IM distortion, aliasing creates new sum and difference frequencies not harmonically related to the original frequencies, so it can be unpleasant and irritating to hear if it’s loud enough. Fortunately, in all modern digital gear, aliasing is so low in level that it’s rarely if ever audible. Aliasing artifacts are sometimes called “birdies” because difference frequencies that fall in the 510 KHz range change pitch in step with the music, which sounds a little like birds chirping. An audio file letting you hear what aliasing sounds like is in Chapter 3.

Transient intermodulation distortion (TIM) is a specific type of distortion that appears only in the presence of transients—sounds that increase quickly in volume such as snare drums, wood blocks, claves, or other percussive instruments. This type of distortion may not show up in a standard distortion test using static sine waves, but it’s revealed easily on an oscilloscope connected to the device’s output when using an impulse-type test signal such as a pulse wave.

TIM will also show up as a residual in a null test when passing transient material. Negative feedback is applied in amplifiers to reduce distortion by sending a portion of the output back to the input with the polarity reversed. TIM occurs when stray circuit capacitance delays the feedback, preventing it from getting back to the input quickly enough to counter a very rapid change in input level. In that case the output can distort briefly. However, modern amplifier designs include a low-pass filter at the input to limit transients to the audible range, which effectively solves this problem.

Time-based errors are those that affect pitch and tempo. When playing an LP record whose hole is not perfectly centered, you’ll hear the pitch rise and fall with each revolution. This is called wow. The pitch instability of analog tape recorders is called flutter. Unlike the slow, once per revolution pitch change of wow, flutter is much faster and adds a warbling effect.

Digital recorders and sound cards have a type of timing error called jitter, but the pitch deviations are so rapid they instead manifest as added noise. With all modern digital audio gear, jitter is so soft compared to the music that it’s almost always inaudible.

The last type of time-based error is phase shift, but this too is inaudible, even in relatively large amounts, unless the amount of phase shift is different in the left and right channels. In that case the result can be an unnaturally wide sound whose location is difficult to identify.

Room acoustics could be considered an additional audio parameter, but it really isn’t. When strong enough, acoustic reflections from nearby boundaries create the comb filtered frequency response described in Chapter 1. This happens when reflected sound waves combine in the air with the original sound and with other reflections, enhancing some frequencies while canceling others.

Room reflections also create audible echoes, reverb, and resonance. In an acoustics context, resonance is often called modal ringing at bass frequencies, or flutter echo at midrange and treble frequencies. But all of these are time-based phenomena that occur outside the equipment, so they don’t warrant their own category.

Another aspect of equipment quality is channel imbalance, where the left and right channels are amplified by different amounts. I consider this to be a “manufacturing defect” caused by an internal trimmer resistor that’s set incorrectly, or one or more fixed resistors that are out of tolerance. But this isn’t really an audio parameter either, because the audio quality is not affected, only its volume level.

The preceding four parameter categories encompass everything that affects the fidelity of audio equipment. If a device’s noise and distortion are too soft to hear, with a response that’s sufficiently uniform over the full range of audible frequencies, and all time-based errors are too small to hear, then that device is considered audibly transparent to music and other sound passing through it. In this context, a device that is transparent means you will not hear a change in quality after audio has passed through it, even if small differences could be measured.

For this reason, when describing audible coloration, it makes sense to use only words that represent what is actually affected. It makes no sense to say a power amplifier possesses “a pleasant bloom” or has a “forward” sound when “2 dB boost at 5 KHz” is much more accurate and leaves no room for misinterpretation.

Chapter 1 explained the concept of resonance, which encompasses both frequency and time-based effects. Resonance is not so much a parameter as it is a property, but it’s worth repeating here. Resonance mostly affects mechanical transducers—loudspeakers and microphones—that, being mechanical devices, must physically vibrate. Resonance adds a boost at some frequency and also continues a sound’s duration over time after the source has stopped. Resonance in electrical circuits generally affects only one frequency, but resonances in rooms occur at multiple frequencies related to the spacing between opposing surfaces. These topics will be examined in more depth in the sections that cover transducers and room acoustics.

When assessing frequency response and distortion, the finest loudspeakers in the world are far worse than even budget electronic device. However, clarity and stereo imaging are greatly affected by room acoustics. Any room you put the speakers in will exaggerate their response errors further, and reflections that are not absorbed will reduce clarity. Without question, the room you listen in has much more effect on sound quality than any electronic device.

However, the main point is that measuring these four basic parameters is the correct way to assess the quality of amplifiers, preamps, sound cards, loudspeakers, microphones, and every other type of audio equipment. Of course, to make an informed decision, you need all of the relevant specs, which leads us to the following.

“The Audio Expert” by Ethan Winer, published by Focal Press (ISBN: 9780240821009), is available here.

Return to articleReturn to article
The Audio Expert: Audio Fidelity, Measurements, And Myths—Part 1