Libraries
 

Contact Information
  
last updated: 10/22/2010

The Match of Perceptual & Acoustic Features..

Introduction and Methodology

     Studies in timbre perception have historically proceeded in two directions: either they have attempted to identify salient features of an acoustic signal that translate into the perception of a specific instrumental tone quality; or they have attempted to connect verbal descriptions of instrumental timbre with some feature(s) of the acoustic signal. Both approaches have yielded results indicating that the phenomenon of timbre - both in perception and description - is multi-dimensional, with a number of factors interacting to produce the exact tone quality that is perceived or described by a listener. The standard definition of timbre in books of auditory perception, in fact, generally describes the phenomenon by negative exclusion, specifying timbre to be all qualities of a tone - with the exception of pitch and loudness - that differentiate one tone from another (ANSI).


     This report describes a project that has two objectives. First, it explores the electric guitar timbre of a genre of popular music known as heavy metal. The concept of "heaviness" is applied by members of the heavy metal subculture to a variety of timbres (bass timbres, drum timbres, vocal timbres) and musical structures (melodic motifs, harmonic approaches) in their music, and heaviness is the defining feature of the genre. While the term "heavy" functions in a variety of ways, it is most commonly used to describe guitar timbres (Berger 1995, Walser 1993), and all other uses of the term are metaphoric extensions of this primary use. The project's first goal is to understand some of the acoustic features that correlate with the perception of heaviness.


     The project's second goal is to demonstrate a new approach for studying timbre. In particular, it expands upon the second approach to timbre research identified above, the linking of a verbal description of tone quality with acoustic features. Previous efforts have proceeded from a consensus by listeners as to the adjectives appropriate to a particular timbre. The methodology demonstrated in this project, however, begins with agreement by specialists in heavy metal music that a particular timbral quality in the music has changed in a specific way over time. In other words, rather than searching for agreement among listeners that a specific sound is characterized by descriptor X, to which the acoustic elements A, B, and C can be correlated, this project takes an historic perspective. Specifically, it is commonly held amongst heavy metal subculturalists that the characteristic timbral quality of the genre has changed since its inception in the early 1970s, becoming heavier and heavier over time, with specific incremental stages marked by prototypical songs (Berger, 1995); as a result, these songs can be analyzed for some acoustic property, or a constellation of properties, that changes in the same direction and increments specified by the listeners.


     The theory informing this methodology is the following. There are two major problems in the effort to correlate adjectives describing timbre with acoustic qualities that elicit those adjectives. One has already been mentioned - the acoustic multidimensionality of timbre. Unlike the sensation of pitch can be fairly directly correlated with the acoustic property of frequency, or the sensation of loudness which correlates to the acoustic property of amplitude, the sensation of timbre is the result of multiple interacting acoustic factors. The second is that, unlike pitch or loudness, there is no relative scale for timbre. That is, timbre does not progress in regular perceptual increments, with all possible timbres arranged along a single trajectory, as pitch moves by regular intervals through an octave, or as loudness moves from soft to loud by perceived increments. However, in a given context - a single piece or genre of music - a listener can distinguish greater or lesser degrees of a given timbre (here, the tone quality is darker than here; now the tone quality is brighter than before). This is especially true if other changes occurring at the same time are easily identifiable as factors other than timbre.


      Many features of heavy metal have remained the same over time. What listeners specify as the quality of "heaviness", however, has been observed to increase incrementally over the genre's history. In its initial stages (early and mid 1970's), heavy metal and commercial hard rock were not fully distinct categories. By the mid to late 1970's, metal was understood as a distinct genre, and throughout the eighties, various hybrids of heavy metal and hard core punk emerged in Britain and the United States. Each new group of bands was considered to have a heavier guitar sound. The early nineties saw the increased popularity of a mid-eighties sub-genre of metal referred to as death metal. The guitar timbres associated with death metal are now found throughout the genre and have spread into more mainstream varieties of popular music as well. In sum, each stage of the history of metal--classic metal 1970's metal, various interactions with punk music in the 1980's, contemporary death metal and post-death metal developments--saw heavier and heavier guitar sounds. The utility of such a delineation of timbre development for the present study is that it not only provides a consensual description of a timbral quality in a genre for which timbre is primary, but also specifies that this quality heightens or intensifies over the course of the genre's history.


     We begin with comments universally accepted among experienced heavy metal subculturalists that the genre in general and guitar textures in particular have become "heavier" over time. Applying previous approaches to the study of timbre, a researcher interested in "heaviness" might proceed as follows: 1) given the adjective heavy, identify one or more acoustic elements tentatively responsible for the sensation of heaviness; 2) present listeners with representative sounds from the genre, but artificially reduced of the identified acoustic elements; 3) ask listeners if they find the resulting sound to be more or less heavy than the original sound; 4) if listeners find the altered sound less heavy than the unaltered sound, then the elements removed from the altered sound can be considered responsible for the sensation of heaviness. The problems with this procedure are many and complex, especially its artificial context which is fiercely disturbing to ethnomusicologists contemplating experimental research. What we have done instead is to select songs commonly associated with this well-known pattern of perceptual change, and analyzed the songs for acoustic changes that correspond to the perceptual change reported by listeners.


Choice of Samples

     The present research is intended as a pilot, a preliminary test of the relevance of consensual perception of timbre change to similarly changing acoustic patterns. For this test, we sought musical examples from the earliest and latest periods in the history of heavy metal music. Several factors limited our selection of recorded samples. Our methodology dictated that we use pre-recorded samples rather than ones recorded in the laboratory, but this constraint introduced a difficulty. Sound analysis is best applied to tones from single instruments, while most recordings involve an entire band playing simultaneously. The cost of commercial recordings and other contingent factors also influenced the examples chosen. In the end, these strictures limited us to four main samples and two additional samples, each an example of isolated guitar playing in a similar pitch range of the instrument. While the sample size was constrained by these various factors, the bands were selected to provide unequivocally typical timbres of their period. Future research might expand on this study in two ways: first, it would employ more samples to indicate the acoustic correlates of the perceptual trends in perception of heaviness in timbres; and second, it would also employ samples from the middle period (the 1980's) to verify the conclusions about the incremental and progressive changes that occurred from heavy metal's origins to the present.


     Introductory passages from Black Sabbath's "Paranoid" (1971) and Judas Priest's "Tyrant" (1976) were chosen as representative samples of 1970's era metal. "In Love" by Grave (1991), a contemporary death metal band, and "Wages of Sin" by Winter's Bane (1993), a contemporary progressive metal outfit with a typically 1990's guitar texture, were selected as examples of the present era. The main samples used were all "power chords" (chords composed of a root, fifth, and octave in the lowest octave of the guitar's range); the power chord is the most common harmonic material used in metal and the typical vehicle for displays of timbral heaviness. Two additional samples of unmuted, single note runs were used to examine dynamic envelop; a short passage from Judas Priest's "Tyrant" was used to represent the earlier period, while a short passage from Winter's Bane was used to illustrate the later period. A comprehensive scanning of spectral representations of these samples led to our choice of acoustic noise - its quality, intensity, and frequency range - as primary candidate for the acoustic features resulting in the percept of "heaviness".


Results

Location of Energy.

     Characteristic of heavy metal music across the 20 years of its history is an immense amount of acoustic noise contributing to its overall timbre. However, examination of spectrograms of the samples reveals the frequency range of both harmonics and noise to cut off at a lower point for samples from the 1970's era than for samples from the 1990's era. For the early period, the Black Sabbath example (henceforth BS, figure A), shows an abrupt reduction of energy above 4.5Khz, with a small amount of additional energy around 6.3Khz; in the Judas Priest example (henceforth JP, figure B), the energy stops abruptly around 5.8Khz. From the late period, the Grave example (henceforth G, figure C), shows energy through 7.5Khz, and in the Winter's Bane example (henceforth WB, figure D) significant energy up to 14Khz, a range beyond the capabilities of most acoustic instruments. Further, both periodic and aperiodic content of the earlier examples stops abruptly at their end points, while all energy in the later examples seems to fade into softer and softer formant-like bands.


     Perhaps more important is a band of energy, with high noise to harmonic content, in the low or midrange of the spectra of all samples; this formant occurs lower in the earlier examples than in the later examples. In the BS example (figure A), the formant runs from nearly the bottom of the spectrum to 800 Hz and is terminated by a sharp notch from 800-1500 Hz. In the JP example (figure B) the formant runs from nearly the bottom of the spectrum to 900 Hz and is terminated by a notch from 900-1700 Hz. The later examples show a rather different pattern. The G and WB examples (figure C and D) show far less energy below 1000 Hz relative to the rest of their spectra than do the early period examples. In the G example, a noise formant exists from about 1.7-4kHz; a notch terminates this noise formant at its top end, from 4-4.5kHz. In the WB example, a noise formant is located from 2-6kHz; a notch terminates this formant at its bottom end, from 1.5-2kHz. In sum, earlier examples have a noise formant below 1.5kHz, and the later examples have a noise formant above 1.5kHz.


Dynamic-Temporal Envelop.

      The dynamic-temporal envelop demonstrates a signal's amplitude change over time. To examine the difference in the envelops of early and late guitar timbres, we sampled isolated examples of unmuted, single note runs. Guitars (and other impulsive instruments, see below) show a characteristic amplitude spike at the beginning of a note and a characteristic decrease in amplitude at the decay end of a note. These features are greatly lessened in both the early period JP and the later WB examples, with attacks still partially visible in the JP example, and virtually nonexistent in the WB example. Figure G shows a spectrogram of three separate attacks in the JP song. While much less evident than the attack of an undistorted guitar, the attacks are still slightly visible at 650 and 850 seconds. Figure H plots amplitude (measured in volts) against time for the same sample and illustrates the same attacks. Figure F illustrates the transition from one note to the next in the WB sample; the transition occurs at about 2200 ms, but the occurrence of a new note is noticeably unmarked by an visible attack; the voltage plot in figure E illustrates the same information. In sum, both examples have a flatter dynamic envelop than one would expect in a guitar, but the later example is far flatter.

Figure E. Dynamic envelop (voltage x time) of sample from Winter's Bane "Wages of Sin", showing transition at about 2200 ms from one note to the next, with little increase in energy at points of attack. Figure F. Spectrogram of sample from Winter's Bane "Wages of Sin", showing transition at about 2200 ms from one note to the next, with little visible change at points of attack.

 

       
Tabular Summary of Results
       
Feature Early Late Results
Heavier timbres correlate with:
       
Energy cuts off BS: above 4.5kHz. (Fig. A)
JP: above 5.5kHz (Fig. B)
G: above 8kHz (Fig. C)
WB: above 8.5kHz (Fig. D)
More high frequency energy.
       
High end energy Ends abruptly (Fig. A and B) Fades off in weaker and formant bands (Fig. C and D) A gradual fading of high end energy.
       
Noise Formant Location BS: Formant, 0-.8kHz Notch, .8-1.5kHz
JP: Formant, 0-.9kHz
Notch, .9-1.7kHz
G: Formant, 1.7-4kHz
Notch, 4-4.5kHz
WB: Formant, 2-6kHz
Notch, 1.5-2kHz
Formant of noise and harmonics above, rather than below, 1.5kHz.
       
Dynamic envelop Weakened but still visible attacks. (Fig. G and H)
WB. No attack is visible (Fig. E and F) Flatter dynamic envelop.


Discussion

Energy placement and noise.

     The results of our analysis provide insight as to specific qualities of heavy metal timbre, and also as to the general connection between acoustic and perceived events. Our search for acoustic elements in target examples that change over time identified the prominent noise component of the timbre to be instrumental in provoking the sense of "heaviness" for which the genre is named; this result is consistent with the belief of metal subculturalists themselves that noisy textures are heavier. More interesting, however, in regard to the relationship between the acoustic and perceived world is the fact that the quality described by informants as becoming "heavier and heavier" over time, translated acoustically into a frequential difference as well as an intensity difference.


     The puzzle, in other words, was this: informants affirm that they hear a quality X, which not only defines the genre that contains it, but which marks the historical development of the genre; that is, the genre demonstrates greater X--an increase in Xness--over time. If X is "brightness" (presumably a timbral quality), then over time, the music's timbre becomes brighter; if X is "syncopation" (presumably a rhythmic quality), then over time the music's rhythm becomes more syncopated; the nature of the pattern, the kind of augmentation described by "increased X" depends entirely on the identity of X. Thus, if the intention is to discover an acoustic correlate to a perceived quality by applying listener's descriptions of a change in that quality, not only is the quality unknown, but the nature of its change is unknown. The researcher must therefore isolate acoustic features that 1) together provoke some perceived quality and 2) intensify together according to the various parameters implicated in their configuration.


     Since informants described the perceived quality of heaviness as increasing, we expected to find an acoustic conglomerate of features whose particular timbral distinctiveness would be emphasized--made more distinctive--over time. If the quality of "heavy" translates into the presence of acoustic noise, then heavier must translate into more noise. But what is more noise? is it louder noise? is it broader band noise? Rather than a simple increase in the noise to harmonic ratio of the sound--that is, rather than becoming louder relative to the non-noisy elements of the sound over time--the noise in heavy guitar timbres changed frequency location as well. Acoustic noise provokes a sensation called "heavy" by listeners, but an increase in that sensation is not simply more noise or louder noise, but--perhaps more prominently--noise that is different in frequency location relative to other elements.


     The correlation of an intensified sensation of a timbral quality with a frequency change is significant in regard to efforts to connect listener descriptions with acoustic structures. In "The Body in the Music", (1991, 123-4), Robert Walser has observed that one of the main effects of distortion in power chords is the expansion of the guitar's perceived frequency spectrum. Walser reports that, on the one hand, the distortion in power chords produces "resultant tones," the perceived impression of energy below the acoustic fundamental of the vibrating strings; on the other hand, the distortion strengthens the upper harmonics of the signal. With respect to the upper range of the frequency spectrum, our research confirms that distortion strengthens the upper harmonics of the guitar's frequency range; further, we observe that timbres perceived as heavier have greatly increased upper harmonics compared to those perceived as lighter. The situation with respect to the lower part of the frequency range is more complex. Resultant tones are a perceptual--rather than acoustic--feature that would not be depicted in the spectrograms of the guitar samples. While resultant tones are not clearly audible in the samples we have examined, the results of our research show that the perceptual quality of heaviness is related to a change in the location of energy in the frequency spectrum. The correlation of the simple linear gradation of the perceptual heaviness with the complex change in size and location of the noise formants is one important result of this research; investigation of this correlation clearly requires further research.


Dynamic Envelop.

     Sharp attacks and marked decays are associated with tones from impulsively stimulated instruments, whereas tones from instruments excited by a constant source of energy to their main vibrating element, usually through bowing or blowing, are characterized by much flatter dynamic envelops. Guitarists in heavy metal and other kinds of popular music prize distortion, for its ability not just to add noise to the instruments timbre, but also to add sustain. Originating in pre-amplifiers, amplifiers, external effect devices, or speakers, distortion allows the guitar player to let notes ring almost indefinitely, a feature achieved without distortion by devices called compressors. But the added sustain that distortion provides has other consequences as well; attacks are flattened, and the final release of the note is foreshortened considerably. In short, distortion simulates the conversion of the guitar from an impulsive to a sustained or driven instrument, and this transformation may be part of the acoustic correlate to the perceptual experience of heaviness. Walser has observed (1993) that the seemingly infinite sustain that distortion provides serves as an icon of power and strength, two values critical to heavy metal's broader ideology. Our research confirms Walser's conclusions and shows that the flattened dynamic envelop is not only correlated with the perceptual quality of heaviness but also that increases in the perceptual quality of heaviness are correlated with flatter and flatter envelops.



Bibliography


Walser, Robert. 1991. "The Body in the Music: Epistemology and Musical Semiotics." College Music Symposium 31:117-124.
Walser, Robert. 1993. Running with the Devil: Power, Gender and Madness in Heavy Metal Music. Hanover, New Hampshire: University Press of New England.
Weinstein, Deena. 1991. Heavy Metal: A Cultural Sociology. New York: Lexington Books.

Discography

Grave. 1991. "In Love". Into the Grave. Century Media CM 7721-4.
Winter's Bane. 1993. "Wages of Sin". Heart of A Killer. Massacre Records MASS CD 013.
Black Sabbath. 1971. "Paranoid". Paranoid. Warner Brothers 3104-2.
Judas Priest. 1976. "Tyrant". Sad Wings of Destiny. RCA AYK 1-4747.
Judas Priest. 1976. "Victim of Changes". Sad Wings of Destiny. RCA AYK 1-4747.



last updated: 10/22/2010