ASA PRESSROOM

Acoustical Society of America
138th Meeting Lay Language Papers


[ Lay Language Paper Index | Press Room ]


Local Performance Recording/Reproduction:
Application to a String Quartet

William M. Hartmann, hartmann@pa.msu.edu
517-355-5202
Zachary A. Constan
Department of Physics and Astronomy
Michigan State University
East Lansing, MI, 48824

Popular version of paper 2pAA4
Presented Tuesday afternoon, November 2, 1999
138th ASA Meeting, Columbus, Ohio

This talk describes and demonstrates a new method of sound recording and reproduction called "Local Performance Recording/Reproduction (LPR/R)." It is fundamentally different from traditional multichannel recording techniques.

TRADITIONAL RECORDING:
A goal of traditional sound recording, stereophonic or surround sound, is to transport the listener into the environment of a musical performance. Ideally, the sound field at the ears of the listener resembles the sound field of a live performance, including ambiance cues to the acoustical character of a performance space.

LPR/R:
The goal of the local performance recording and reproduction technique (LPR/R) is the reverse. It attempts to transport the music into the environment of the listener. Both the recording and the reproduction steps are different. The LPR/R technique requires at least one separate recording channel for each musical instrument. Recorded channels contain no crosstalk among instruments and no reflected sound from the room surfaces. On reproduction, the channels from different instruments are similarly kept separate; they are never electrically mixed. Instead, the channels for each individual instrument are reproduced by a dedicated array of loudspeaker drivers, attempting to simulate the radiation pattern of the instrument.

The LPR/R technique is effective for small ensembles. An eight-channel application of the technique to a Mozart string quartet is described and demonstrated.

FEATURES OF LPR/R:
Each instrument is recorded and reproduced individually. This procedure results in the following features:

(1) This is no intermodulation (IM) distortion between the tones of different instruments. Because the tones of a musical instrument tend to be harmonic, the only consequence of the inevitable distortion in loudspeakers is a slight change in coloration for the instrument. By contrast, IM distortion between different instrument tones is a deficiency of all other means of sound recording/reproduction. It is acknowledged by everyone that this is the worst kind of distortion. Eliminating it with the LPR/R technique is a significant benefit.

(2) For better or worse, the reflections and reverberation experienced by the final listener are those of his or her own room. This is important in creating the impression that the music originates from within the room.

(3) Because the reproduction level of each instrument of an ensemble can be individually controlled, a music instructor can highlight a particular musical instrument in an ensemble for educational purposes. Similarly, the tone color of each instrument can be varied to taste. For instance, in listening to a simulated string quartet, the listener could elect to give the second violin a darker tone color to exaggerate the difference between it and the first violin.

(4) Because any instrument of the ensemble can be individually shut off entirely, LPR/R provides the ultimate "music-minus-one" system. The performer feels that the other musicians of the ensemble are with him, and around him, in the same room.

(5) Because each instrument is recorded separately, editing of recordings and processing of individual voices is facilitated. Errors by one musician can be corrected without the participation of the other musicians.

(6) Because their intended function is limited, loudspeakers can be optimized for their particular functions. There is no need for a large low-frequency driver (woofer) in the system that is dedicated to a flute. Dedicating loudspeaker systems controls the cost of multichannel ensembles.

IMAGING: PRESENT STEREOPHONY VS LPR/R:
In an important sense LPR/R is the opposite of the present stereophonic recording practice. Stereophonic recording attempts to bring the listener into the environment of the music. By contrast, LPR/R attempts to bring the musical performance into the environment of the listener.

Present stereophonic practice creates a musical environment by including room information, early reflections and reverberation. Recording engineers pay close attention to the room and the location of microphones when they record ensembles. When the original recording is devoid of room information (because it is made electronically or by tight-miking techniques) room information is added electronically through reverberation lines and ambiance synthesizers.

Present stereophonic practice sometimes attempts to localize sound images, but localization is psychoacoustically fragile, depending on the loudspeakers, listening room, and listener position used by the ultimate consumer. Therefore, the principal function of stereo is to delocalize the sounds from the loudspeaker positions themselves and to provide a broadened image.

An important feature of LPR/R is its imaging capability. The listener perceives the individual instruments (voices) to be spatially compact, well-localized in azimuth, elevation and distance. That is because the technique physically recreates the radiation pattern of the ensemble within the listener's environment. Unlike conventional stereophonic sound reproduction, or contemporary surround sound techniques, the LPR/R does not require the listener to be in any one place. The listener can walk entirely around the synthesized performing ensemble. In fact, the imaging seems to be best when the listener moves.

APPLICATIONS OF LPR/R:
The LPR/R technique requires a minimum of one recorded channel and one transducer in the listening environment for each instrument (voice) in the ensemble. The cost and space requirements grow approximately linearly with the size of the ensemble. Ensembles larger than an octet would seem cumbersome at this time.

Application to string quartets is a natural. (1) Technically it works well. (2) There is a vast and wonderful literature for string quartets. (3) There are many performers and devoted fans of string quartets.

Conventional two-channel distribution of sound, via stereo broadcasting or compact discs or tape, is inadequate for LPR/R. The LPR/R technique depends on new technology (not far off) for multichannel distribution.

REDUCTION TO PRACTICE:
The concept of the LPR/R technique is probably more significant than any particular application. Details of the application depend on the musical instruments that are recorded and reproduced. Our reduction to practice (MSU Physics, 1997-1999) applies the technique to a string quartet.

Recording: Contact pickups are used on the four instruments. In that way the recording achieves the mutual separation of the instrument sounds and the freedom from reverberation required by the technique, while allowing the musicians to play together in the normal way. This was felt to be a great advantage over a non-simultaneous recording technique. A problem with contact pickups is that they lead to a rather uninteresting sound. In order to obtain a final result with an interesting sound pattern, two contact pickups are used for each instrument, requiring an eight channel recording for the four instruments. The choice of the locations for multiple pickups (transducers) was determined by measuring the cross-correlation function between transducers taken in pairs as measured in different frequency bands.

Reproduction: Because the loudspeaker systems for individual instruments are physically separate and can be placed appropriately in the listener's room, the radiation pattern of the ensemble can approximate the pattern of instruments in space. A further goal of the reproduction technique is that each loudspeaker system should simulate the radiation pattern of the instrument to which it is dedicated. Our present technique has three components:

(1) Overall frequency response: The overall frequency response of the simulation, measured in one-third-octave bands, should equal that of the original instruments. That goal is achieved by the following procedure: At the recording session, each musician plays scales in a reverberant environment. Recordings are made with contact transducers as usual and, using a separate recorder, with acoustical microphones. In constructing the reproducing system, the loudspeakers are adjusted so that when they reproduce the signals from the contact transducers, the long-term spectrum measured with the same acoustical microphones in the same reverberant environment matches the original recordings. It is expected that perceptually important formant structures in the real instruments will be captured by the third-octave matching technique.

(2) Asymmetrical frequency response - overall structure: The overall directional frequency response of musical instruments has been measured in anechoic rooms by many workers. For example, Jurgen Meyer has measured the angular dependence of the frequency response for many orchestral instruments including violin, viola, and cello. These responses appear in his 1978 textbook, "Acoustics and the Performance of Music."

The LPR/R technique attempts to match measured angular dependence with loudspeakers by two techniques. First, the frequency dependence of the radiation from front and back is approximated by using separate loudspeaker drivers for front and back emission and by filtering the inputs to each driver. Second, the polar pattern of radiation is approximated by using drivers with a piston diameter that reproduces the low-frequency lobe in the forward direction. For example, at an angle of 90 degrees the radiation from a viola is down 3 dB at a frequency of 1000 Hz. According to a well-known theory for the radiation of a piston in an infinite baffle, a polar pattern with that characteristic requires a piston diameter of about 22 cm.

(3) Asymmetrical frequency response - fine structure: The fine structure of the radiation pattern of a musical instrument is complicated. For a violin, the fine structure is even different for each different violin. The result of the fine structure is that when the musician plays changing notes, the different high-frequency harmonics are radiated into directions that change dramatically. This effect lends interest to the sound of the instrument; the tone is lively.

The implementation of the LPR/R technique to the string quartet does not attempt to reproduce the fine structure of any particular instrument. What is thought to be important is simply that some complicated fine structure be present. For each instrument of the string quartet, the radiation from the top of the instrument comes from two loudspeakers, driven by a weighted mixture of bridge and f-hole signals, with possible inversion. The resulting interference pattern leads to fine structure.