As mentioned in my last post, I recently visited Volker Mellert in Oldenburg. During our oral-history interview, he told me about the three dummy head projects he had been involved. Mellert started his dummy head research around 1968 at the III. Physical Institute at the University of Göttingen. The III. Physical Institute, under the direction of Erwin Meyer, was specialized in studying all kinds of wave phenomena from microwaves to concert hall acoustics. Mellert studied physics in Göttingen with no special interest in acoustics, but he particularly enjoyed the practical training the III. Institute offered. Following this practicum, Mellert received the offer to write his diploma thesis about stereo reproduction systems – we’ll come back to this later.

Roughly at the same time, as Mellert started his diploma project, two of his senior colleagues, Peter Damaske and Bernhard Wagener, studied cross-correlations of ear signals. The initial observation was that for sound directions “front” and “back” human spatial hearing could not get directional cues from time and sound pressure differences; their assumption was that human hearing uses direction specific sound variations to discriminate “front” and “back”. Instead of using probe microphones to measure ear signals of human test subjects, Damaske and Wagener decided to use a dummy head. For this purpose, they acquired a mannequin’s head from a nearby department store. Because the mannequin’s ears had no proper pinnae, Damaske and Wagener had to carve them out themselves (in the picture below you can see the white carved out part of the mannequin’s ears). They approximated progressively the dummy head’s ear signals to measurements of human ear signals they obtained from the literature. For measurements they inserted two curved probe-tubes and coupled them with Sennheiser MD 321 probe microphones (see pictures below). Damaske and Wagener used their dummy head for localization experiments in the median plane. In an anechoic chamber, seven test persons listened, first with their own ears, and then through the ears of the dummy head to a human speaker reading continuous text from eleven different directions. Test persons had to judge directions and results of the two test series were compared. The result was that test persons had greater difficulties to judge the sound direction (in particular, from the direction “back”) through the dummy head’s ears. Still, the experiments showed that a dummy head could be used for such localization tests (Damaske/Wagener 1969).

kuko damaske & wagener

Dummy head of Damaske and Wagener; the light grey spot is the carved out pinna. At the bottom you can see the probe microphones. Source: Damaske/Wagener 1969.

In 1970, an improved version of the dummy head was also used for binaural recordings (Damaske 1971): e.g. of a jazz concert and a choir of 120 singers. Damaske, Wagener and Mellert realized that these recordings were “very impressive and lively” although their dummy head was not built for music recordings. The probe microphones had an unfavorable signal-to-noise ratio compared to studio microphones as their initial aim was to find cross-correlations of ear signals to better understand human spatial hearing and not to produce high-fidelity recordings. However, favorable for the reproduction of binaural recordings was the stereo reproduction system, Mellert constructed in his diploma project. In close cooperation with Peter Damaske, he built a system to reproduce head-related stereophony with two loudspeakers. Because dummy head signals are supposed to receive only one ear (here: left loudspeaker signal only the left ear, and the right loudspeaker signal only the right ear), Mellert constructed a 90° filter to compensate for the acoustical cross talk of conventional stereo reproduction (Damaske/Mellert 1969). Damaske concluded: “This transmissions system was tested for its directional fidelity with many test persons. Localization is nearly ideal for the entire horizontal plane including front-back discrimination and is partly correct even for the median plane” (Damaske 1971). Unfortunately, this system only worked in an anechoic environment.

 

2L

Schematic drawing of Mellert’s head-related stereo reproduction system. Source: Kuhl/Plantz 1975.

The second dummy head was used for new studies in 1973/74. The new director of the III. Institute, Manfred Schroeder, obtained a research grant from the German Science Foundation to compare the acoustical quality of 22 European and North American concert halls. Based on previous works of Damaske, Wagener and Mellert, Karl Friedrich Siebrasse and Dieter Gottlob built a new dummy head for this study. They used an improved dummy head which had been developed by Mellert in 1970/71: its acoustical properties should resemble those of an “average” human listener. Mellert had measured hearing thresholds of 17 test persons and then calculated the correlated structures of all individual curves. Averaging the correlated structures resulted in a frequency curve which was used to built pinnae and eardrums of the dummy head. The pinnae were actually casted from the ears of a test subject whose pinnae matched the average geometrical size of the ears of 30 persons (Mellert 1972). Siebrasse and Gottlob used this improved dummy head to record music in different concert halls “and then to play these recordings back at a convenient time and place, allowing for rapid switching between different halls.” To ensure comparability of recordings they radiated a test tape of a two-track recording of the 4th movement of Mozart’s Jupiter Symphony made by the English Chamber Orchestra from the stage (instead of recording a live orchestra). Binaural recordings were then used for subjective evaluations of the different concert halls; furthermore, subjective judgments were employed to find correlations with objective properties such as volume or width of the hall (Schroeder/Gottlob/Siebrasse 1974).

kuko siebrasse & gottlob

Siebrasse and Gottlob with (second) dummy head. Source: Schroeder 2004

The third artificial head was built in Oldenburg: in 1974, Mellert was appointed professor for applied physics at the newly founded University of Oldenburg. Together with Reinhard Weber he started to construct a new dummy head. This head was again based on the concept of constructing the “average” listener (the basic idea was already noted in Damaske’s and Mellert’s 1969 publication). Mellert and Weber used data from existing literature to calculate mean values of head size and ear size and form. The actual ears were again taken from a test person who’s ears were closest to the “average” ear. The mould was produced on a CNC machine, this is why the ear has this “weird” steplike geometrical form (see pictures below).

The Oldenburg head was originally equipped with SCHOEPS studio microphones (later it was also equipped with standard microphones from Brüel & Kjær). The studio microphones allowed for high quality recordings. A small number of heads was sold to different automobile manufacturers; they used binaural recordings for sound design and noise control purposes. At least two Oldenburg heads were sold to Delta Acoustic Studio, a music recording studio in Wilster (a small town in Northern Germany). Sound engineer Manfred Schunke, founder of Delta Studio, experimented with dummy head recordings and, in 1978, convinced Lou Reed to use dummy head microphones to produce his next three albums. Schunke used, for example, two Oldenburg heads to record ten Lou reed concerts in New York’s Bottom Line club. Back in Wilster, Schunke and Reed chose the “best”, or most “natural”, binaural recordings for the LP “Live – Take no prisoners”. Dummy head recordings were also used in the production of the studio albums “Street Hassle” (1978) and “The Bells” (1979). Delta also tried to commercialize the Oldenburg head under the label of Delta head – with little to no success (see advertisement below).

delta

Advertisement for “the Delta Head” (end of 70s). Source: Private Archive V. Mellert

To conclude, dummy head research in Göttingen and Oldenburg was driven by two fundamental research interests: a) to better understand human spatial hearing, i.e. to understand which sound properties were important for sound localization, and which weren’t. This research strand was informed by psychoacoustic and neurophysiological findings of that time; b) to better understand how synthetic sound fields could be produced or manipulated to give listeners good spatial impressions. This research strand was driven by improved computer technology of that time.

Sources

Damaske, Peter & Bernhard Wagener (1969). Richtungshörversuche über einen nachgebildeten Kopf. Acustica 21: 30-35.

Damaske, Peter & Volker Mellert (1969). Ein Verfahren zur richtungstreuen Schallabbildung des oberen Halbraumes über zwei Lautsprecher. Acustica 22: 153-162.

Damaske, Peter (1971). Head-Related Two-Channel Stereophony with Loudspeaker Reproduction. The Journal of the Acoustical Society of America 50 (4): 1109-1115.

Kuhl, W. & R. Plantz (1975). Kopfbezogene Stereophonie und andere Arten der Schallübertragung im Vergleich mit dem natürlichen Hören. Rundfunktechnische Mitteilungen 19 (3): 120-132.

Mellert, Volker (1972). Construction of a Dummy Head after New Measurements of Thresholds of Hearing. The Journal of the Acoustical Society of America 51 (4): 1359-1361.

Schroeder, Manfred, Karl Friedrich Siebrasse & Dieter Gottlob (1974). Comparative study of European concert halls: correlation of subjective preference with geometric and acoustic parameters. The Journal of the Acoustical Society of America 56 (4): 1195-1201.

Schroeder, Manfred (2004). Computer Speech: Recognition, Compression, Synthesis. Berlin & Heidelberg: Springer.

Oral-history Interview with Volker Mellert, March 9, 2016.

 

Advertisements