"Human Supervision and Control in Engineering and Music"

Alexander Galembo 

Perception of Musical Instrument  by Performer and Listener  (with application to the piano)

Abstract

Instrument Quality

For many years I was in charge of the acoustical laboratory of the Leningrad Musical Instrument Industrial Corporation in the former Soviet Union. The primary task of the laboratory was to solve acoustical and technical production problems related to the quality of musical instruments made by the Corporation.

The standard approach to our task looked easy: Experienced musicians should tell us in detail which instruments had better tone and better mechanical response. Then we should compare these instruments using physical measurements, find a correspondence between the subjective and objective differences in good vs. poor instruments, and make practical conclusions about technological changes necessary for improving the instruments produced.

After several unsuccessful attempts to follow this logical way of working we came to the suspicion that the main source of our failure was hidden in the subjective information about the quality differences between instruments. This „misleading“ information was, however, received from high-class musicians, who, of course, were sincere in their opinions. An analysis of the psychological conditions under which the musician’s subjective assessment of the quality was made seemed to be a natural step. 

After analyzing our own experiences and many scientific publications, we came to the conclusion that the musician is often put in a situation, in which (s)he is not really able to estimate the quality of the instrument properly. The reason is that the musician is exposed to some hidden but significant psychological factors, influencing the interaction with the instrument, and the judgments will be biased. Some of the most prominent factors are: 

1. The acoustical characteristics of the room are responsible for characteristic changes in the instrument sound. The performer is usually not able to separate the influence of the room but attributes it to the instrument. The same instrument sounds "soft" when played in a large "lively" room, "ringing" in a small empty room, and "short" in an anechoic chamber. During one period, our piano factory received many complains from customers that our upright pianos sounded too bright, „glassy.“ The reason was finally found in a raised noise level in the voicing cabin at the piano factory. 

 When playing a musical instrument, the musician intuitively adapts to the acoustical conditions of the room. For example, the pianist adopts the finger forces applied to the keys to reach the desired tone [1]. Therefore, not only the tone quality but also the mechanical response of the instrument may be perceived as different in different acoustical conditions. 

2. Education and playing experience develop a prejudice, meaning an unconscious preference for (1) an instrument from a renowned maker, (2) old or historical instruments [2], or (3) for instruments similar to the musician's own instrument, etc. For example, the sound of a Stradivarius violin will likely be rated higher than other old Italian violins, once the listeners have been informed about the makers names. In such listening tests a "different" tone quality compared to the established ultimate maker is equivalent to "poorer," almost by definition. A similar situation exists among pianos, with Steinway & Sons as the „standard of excellence.“ 

The most popular method for an "unbiased" evaluation of the tone quality is to let a player perform on the instrument behind an acoustically transparent, but visually opaque curtain. It is necessary to understand that the acoustic transparency of such curtain is not absolute [3]. Further, and much more important, in such presentations the expert listener estimates not only the instrument, but the instrument as a part of a complex including the playing musician [4]. If we restrict the judgment to individual tones, the player's contribution to the tone quality will vary depending on which family the instrument belongs. For instance the player’s influence is weak in pianos and strong in bowed instruments and wind instruments. The player´s influence becomes much stronger when music excerpts are played instead of individual tones. Even for pianos, expert listeners seem to recognize familiar pianists from recordings better than familiar instruments.

A logical way to eliminate the player´s influence is to use a mechanical playing device. However, in this way we loose the information about the adjustment process of the musician to the instrument - the easiness of this adjustment is also one of the many components of instrument quality. Another way is to avoid musical excerpts as stimuli and only use individual tones, chords and scales. This will, however, make the evaluation a quasi-technical rather than an aesthetic exercise. 

There are also other psychological factors affecting the musician's judgments of musical instruments for engineering tasks. Significant differences in terminology used by engineers and musicians, as well as other individual characteristics of the expert may reduce the usefulness of the evaluation results for improving instrument design. 

In particular, one specific psychological factor - originating from the multi-modality of the interaction between the performer and the instrument - is of great importance. This factor is clearly illustrated when the experimenter asks the musician to evaluate the tone quality and the (mechanical) playing comfort separately. This is a typical task, because it should help the piano engineers to address the quality problem at the appropriate level of the design.  It is not well known, neither by experimenters, nor by musicians, that this task is impossible for a musician to execute reliably in normal playing conditions.

The difficulties associated with this type of evaluations were clearly illustrated by the results from an experiment organized at the Leningrad piano factory in the late 70´s [5]. At that time we received much criticism from musicians regarding the poor tone quality of the concert grand pianos made by our factory. Usually this poor quality was contrasted against the high (desired) quality tone of the Steinway concert grands from the Hamburg factory, the most popular piano among professionals in the Soviet Union.  To resolve some doubts about the reliability of the musicians' judgments, we organized a complicated and rather expensive three-step evaluation experiment, with 12 professional pianists as expert players/listeners. All pianists were pedagogues at the Leningrad Conservatory of Music 

Step 1: Free estimation. Three concert grand pianos - Leningrad-made, Hamburg Steinway, and Bechstein were placed on the stage of the concert hall in the conservatory. The experts were asked to play whatever they wanted and to compare the three pianos in three pitch ranges (bass, middle, and treble) with respect to tone quality, dynamic range, and playing comfort. The experts were asked to fill out  forms, using (1) free verbalizations of their impressions, and (2) to rate the pianos by their tone, mechanical, and overall qualities. The last question on the form was whether the expert thought that he would be able to discriminate the instruments by their tone quality only, if presented to tones, chords or scales played on the pianos.

The results of this free estimation were absolutely as expected. The tone of the Steinway piano was judged the best in all registers, followed by Bechstein, and the last was the Leningrad piano. No clearly expressed comments were given about the differences in the playing comfort. All experts found the differences in tone quality so distinct that they had no doubts of that they would be able to discriminate the instruments by listening to played tones, chords and scales.

Step 2: Listening test. Listening tests consisting of separate tones, scales and chords played on the three pianos behind an acoustically transparent curtain were conducted. The listener's tasks were  (1) to identify which of three pianos was played, and (2) to order two stimuli with respect to sound quality. 

In this test, the result was in short that the musicians were not able to determine properly which piano was played, and their preferences in ordering stimuli did not correspond to those expressed in step 1 of the experiment (see Fig. 1, cases 1-4). 

Step 3: Blindfold and kinesthetic tests.  The three pianos were positioned on stage so that the keyboards formed a triangle. A rotating chair was placed in the middle of the triangle. The expert was blindfolded by soft eye covers before seated on the rotating chair. The experimenter rotated the chair and stopped it in a position facing one of the pianos. The task of the subject was to play the piano and to identify which of the three pianos he was playing.  Following, a kinesthetic test was performed, in which the blindfolded expert was  „deafened“ by headphones fed by white noise.

The result of the blindfold and kinesthetic tests were striking - almost all answers were correct (see Fig. 1, cases 5 and 6). Even the subjects were surprised by the results. For us, engineers and acousticians, this meant much. The results of the three steps of the experiment indicated that the quality difference between the three pianos, which was attributed by the experts to the tone quality, could be attributed primarily to the mechanical response, thus dramatically re-focussing the industrial R&D from the tone generation units (hammers, strings, and soundboard) to the key action.

Fig.1:  Identification of pianos by quality. Stimuli for listeners: (1) Chord C-dur in 4 octave,  (2) Chord C-dur in 2 octave,  (3) Sequence c-e-g in 4 octave, (4) Musical fragment on the subject’s choice; 
Stimuli for performers:  (5) Blindfold test (playing with hearing but not seeing), (6) Kinesthetic test (playing without hearing and seeing). The white horizontal dashed line indicates the chance level at 33.3%>

The results of these tests are explainable if the interaction between the performer and instrument is analyzed from a psychological point of view (Fig. 2).

For the pianist, the tone is the auditory feedback of his own activity. The pianist controls the piano via his/her kinesthetic system (arms, hands, and fingers) and perceives a feedback via both kinesthetic and auditory sensory channels. There are good reasons to assume that the pianist's opinions about the „touch“ (the production) and the tone quality (the perception) are intertwined. Our experiments suggest in particular that pianists unconsciously include the kinesthetic feedback when they evaluate timbre, and vice versa. The  performers’ judgments of the tone quality were „infected“ by the impression of the mechanical response due to cross-modal interference, a phenomenon well known from psychology. 
 

Touch

Fig.2:  Psychological conditions of Performer/Instrument/Listener interaction

Let us recall that the performer - instrument interaction is not only a physical, but also a psychological process, and then reformulate the question: Why do some pianists think that the timbre of their individual tones is touch-sensitive? Well, it might be that the performer's judgment of the tone is affected by some touch-sensitive information, which is not available to the listener (14). This seems also to be the case.

First, the pianist hears more details than the listener just because of being closer to the sound source. Particularly attack noises in the key action, that certainly are touch-dependent, will be audible to the player. Second, since the pianist controls the generation of the tone with the key, (s)he has much more exact expectations and thus better tracking of many small details of the complex event named "tone". Particularly, for the pianist, the tone starts physically when the finger contacts the key (mentally even earlier) and includes all possible audible touch-dependent attack elements, as well as the mechanical feedback from the action via the key. The listener's perception of the tone includes only the auditory components. Therefore, for the pianist, the time interval between the finger-key contact and hammer-string contact (between 25 and 80 ms approximately and controlled by touch) is a part of the perceived tone, while the listeners may at best have only a vague idea about that activity during this part of the touch (via the level of the attack noise). 

Further, as shown above, the pianists unconsciously include the kinesthetic information when evaluating the timbre. All this makes not only the tone itself, but rather the pianist’s judgment of the tone, touch-dependent.
 

Melody lead in Piano Performance

It has not been clear, however, whether this melody lead is applied by the pianists deliberately, in a way that it can be controlled independently of other expressive parameters [11].  Another explanation, which has been offered, referred to as „velocity artifact“ [12], deals with the timing characteristics of the piano key action and is more technical than musical.  It says that the melody lead is not the product of a performer’s deliberate strategy, but is just the result of physical processes in the key action, which respond faster to tones played louder.

To solve this controversy, Werner Goebl of the Austrian Research Institute for Artificial Intelligence recently monitored 22 skilled pianist playing pieces by Chopin on a Bösendorfer SE290 grand piano with monitoring of the key and hammer motions. The asynchronies in the finger-key contacts of different voices were related to the asynchronies at hammer-string contact level, i.e. the actual sounding start of the tones [13]. 

The results obtained  indicated that on the average the significant asynchronies, which did exist at hammer-string contact level (melody lead), were not present at the finger-key contact level. This means that the „velocity artifact“ is responsible for most of the melody lead phenomenon. In other words, the pianist has not much choice - in most cases (s)he just exploits the inherent timing characteristics of the piano action for separating the dynamically emphasized melody from other voices in time. 

The three practical situations described above (instrument quality rating, touch and melody lead) give us a clear evidence of that musical performance cannot be fully explained in a straightforward deterministic manner. Due to the non-additive nature of the multi-modal interaction between the performer and the musical instrument, this process represents a complicated mix of deliberately controlled and unconscious elements, leaving much space for the performer’s imagination as well as for further psychological and physical analysis.
 

References

2. Podgorny, T. (1924) „Influence of the age of bowed instruments on musician's judgement (in Russian),“ Musikalnaia Kultura, issue 3.

3. Meyer J. (1977) "Die Problematik der Qualitätsbestimmung bei Musikinstrumenten," Instrumentenbau-Zeitschrift, Heft 2, pp 3-8. 

4. Rimski-Korsakov, A., Diakonov, N. (1952). Musical instruments (Methods of research and calculation), MIR Publishing House, Moscow.

5. Galembo, A. (1982)  "Quality evaluation of musical instruments“ (in Russian) Technical Aesthetics 5, 16-17.

6. Bryan, G. H.(1913)  „Pianoforte touch,“ Nature 91, 246-248.

7. White, W. B. "The human element in piano tone production,"  J.Acoust. Soc. Am 1, 357-365 (1930). 

8. Askenfelt, A. & Jansson, E. (1991) "From touch to string vibrations. II: The motion of the key and hammer," J.Acoust. Soc. Am 88, 52-63.

9. Kornhoof,  G. W. & van der Walt, A. J. (1993): "The influence of touch on piano sound," Proc. Of  SMAC'93, pp. 318-324. 

10. Hartmann, A. (1932). „Untersuchungen über das metrischeVorgalten in musikalischen Interpretationsvarianten,“ Arch. Gesamte Psychologie 84, pp. 103-192.

11. Palmer, C. (1996) „On the assignment of structure in music performance,“ Music Perception 14, 23-56.

12. Repp, B. (1996) „Patterns of note onset asynchronies in expressive piano performance,“ J.Acoust. Soc. Am. 100, 3917-3932.

13. Goebl, W. (2001) „Melody lead in piano performance: Expressive device or artifact?,“ J.Acoust. Soc. Am 110, 563-572.