.

von FLOYD K. BECKER (von BELL Telephone Laboratories) im Mai 1959
* Member of the technical staff, Bell Telephone Laboratories.
.

• Anmerkung : Es kommt nicht so genau aus dem Artikel heraus - aber es geht um die eigentliche Frage, wieviel Kanaltrennung braucht man wirklich für einen "gesunden" Stereo-Effekt ?

.

If stereo broadcasting is to become acceptable, either transmission channel must be capable of presenting a complete signal to the listener not equipped for stereo presentation. This system is completely compatible for use with any two-channel transmission.

The "new sound" of high fidelity stereophonic sound is far from new. At least as early as 1881 a demonstration of an electrical transmission of binaural sound, a close cousin of stereophonic sound, was made at the Paris Opera. That demonstration used two widely spaced microphones connected to a binaural headset.

In 1925, the New Haven radio station, WPAJ, made binaural broadcasts by employing two separate AM transmitters on different wavelengths. Two standard studio microphones placed 7 inches apart originated the binaural signals.

Three microphones were placed before the orchestra, one on each side and one in the center at about 20 feet in front of and 10 feet above the first row of instruments of the orchestra. The microphone outputs were transmitted to Constitution Hall in Washington by three separate electrical circuits specially tailored for the occasion.

At Constitution Hall these transmission lines were connected to power amplifiers. The associated loudspeakers were placed on stage in positions corresponding to the microphones in the Academy of Music in Philadelphia. The comments of those who heard this reproduced concert proclaimed the development of a system with possibilities for even greater emotional appeal than that obtained when listening to the orchestra "live." Much of this reaction was undoubtedly due to the enhanced volume range.

In 1933, the Bell Telephone Laboratories culminated a series of auditory perspective tests in a three-channel system demonstration. Under the auspices of the National Academy of Sciences, a concert of the Philadelphia Orchestra was transmitted from the Academy of Music in Philadelphia to Constitution Hall in Washington. The orchestra was conducted by Associate Conductor Smallens while Dr. Stokowski, the Director, manipulated electrical controls from a box in the rear of Constitution Hall.

In 1936, the Bell Telephone Laboratories produced disc recordings of two-channel stereo. In 1939, Philips of The Netherlands experimented with stereophonic sound reproduction in large auditoriums. In 1941, the Bell Telephone Laboratories demonstrated sound motion picture recording using three-channel stereo.

This chronology of early sterephonic sound is by no means complete. It only attempts to place a few of the salient efforts at stereophonic sound transmission and/or reproduction.

Radio broadcasting of stereophonic sound programs by two separate channels became popular in 1952-4. In various experimental arrangements, the two channels required have been selected from different combinations of AM, FM, and TV, the listener spacing appropriate receivers properly in his home.

Results have been sufficiently favorable that more broadcasters are considering offering stereophonic sound programs. Many such programs have been originated on the national networks.

The major obstacle to vastly increased use of this type of stereophonic broadcasting, however, is the person who listens with only one receiver. If the broadcaster tries for the full stereophonic effect, the sound the single channel listener hears comes from only one of two widely spaced microphone pickups, and he misses a portion of the program.

The effect in many cases is similar to listening to one-half of an orchestra or to one side of a two-way conversation. What the single channel listener does receive is poorly balanced, because of the microphone placement in relation to the sound sources. The broadcaster, in order to protect investment and sponsor revenue, has had to dilute the stereophonic effect in order to preserve satisfactory reception for the single channel listeners.

Most of the effort to produce a compatible stereophonic sound system has been directed at single channel systems. These systems generally comprise a frequency or time multiplexing of the stereophonic channels on a single carrier. Most of these systems are, indeed, compatible with present day single-channel receivers but require additional equipment - other than AM and FM receivers - to produce stereophonic sound.

One solution to the two-channel problem is gained through the use of a compatibility circuit developed at the Bell Telephone Laboratories. This circuit, equally adaptable for a two or three-channel system, depends for success upon a psychoacoustic phenomenon known as the Precedence Effect.

Before discussing this effect, it would be well to review the principles of sound localization. The localization of a sound source with respect to the observer requires three coordinates: the radial distance, the altitude angle, and the azimuth angle. Man's auditory perception of distance seems to be primarily governed by the loudness and ratio of direct to reverberant sound. There is little or no altitude perception.

Azimuth localization is extremely good, and accuracies of about 2 deg. are average. The mechanisms for azimuth detection are (1) phase differences between sound waves at the two ears; (2) differences in the time of arrival of transient sounds; (3) differences in intensity at the two ears due to shadowing by the head, which also has a frequency dependence and will result in an interaural quality difference.

Azimuth localization of pure tones is possible only in areas approximating free-space. In reverberant rooms, even those with optimum reverberation time for music reproduction, the standing wave patterns destroy the sense of directivity. Hence the arrival time and intensity differences of transients assume the predominant roles in determining azimuth in ordinary listening environments.

It is possible to trade loudness differences for arrival time differences. An observer seated before two in-phase loudspeakers which are separated by several feet will gain the impression of a single, centrally located source if the two loudspeakers have the same loudness.

If the loudness of one speaker increases while the other correspondingly decreases, the apparent source will shift toward the more intense loudspeaker. The amount of the apparent shift depends upon the sound intensities at the ears.

If the sound levels are the same in both speakers, but a time delay is introduced in one source, the apparent source will shift toward the underlayed speaker.

• Amerkung : Der Autor unterscheidet in 1959 (damals noch) nicht zwischen Audio-Pegel, Lautstärke und Lautheit seiner beiden Stereo-Signale.

.

Time delays as short as 250 microseconds produce considerable shifts (erhebliche Verschiebungen) in the apparent source. These time delays are of the same magnitude as the time differences of transient arrivals at the two ears for a sound source to the observer's side.

When the time delay in the one speaker system is increased to 2 or 3 milliseconds the delayed source must be intensified until it is ten times more powerful than the undelayed source before the observer will detect it to be of the same loudness! This condition holds while the delay is increased to about 35 milliseconds.

In the neighborhood of 35 milliseconds time delay, the observer begins to detect a distinct echo. It is the reaction of the azimuth localization mechanism in the region of 1 to 30 milliseconds that is called the Precedence Effect.

In this region, the localization of a sound source is determined by the direction of the first arriving sound. The later arriving echoes are virtually disregarded. This may at first seem an amazing phenomenon but a closer examination discloses that it is at least an every day experience.

In the average indoor environment, the bulk of the sound power reaching a listener arrives by way of reflections or echoes. Yet, in this same environment the listener has no difficulty localizing the sound source.

Now to turn to Figs. 1 and 2 and an explanation of this form of a compatible stereophonic sound system. Referring first to Fig. 1, the circuits between the microphone pickups and their corresponding radio or TV transmitters are cross connected through two delay lines, each with its own buffer amplifier.

Because of these cross connections, music or voice signals from the left microphone are transmitted directly to the left loudspeaker in the listener's home, while the same signal is slightly delayed before reaching the speaker to his right.

The stereo listener will hear the sound as if it came only from the left loudspeaker because of the Precedence Effect. Conversely, the sound from the right microphone goes direct to the right speaker, but is delayed before reaching the left speaker, and is therefore unheard in the left speaker. Thus, the stereo listener localizes the sound he hears as coming direct from each of his two speakers, and full stereophonic effect is maintained.

However, monophonic reception is completely compatible with this, since a listener to each single channel hears the total sound from both microphones in a balanced reproduction. The slight delay of one signal does not affect his reception at all.

The three-channel system of Fig. 2 operates in a similar manner. The direct signal travelling only in the primary channels while a time delayed replica of the direct signal is added to the alternate channels in order to achieve compatibility.

Bildunterschriften:
Fig. 1. (below) Block schematic of Precedence system for two channels. Fig. 2 (right) The same system may be expanded to permit transmission of three compatible channels.
.

Results of two-channel subjective listening tests with musical material indicate a preferred time delay of about 10 milliseconds with the intensity of the delayed signal equal to the direct signal.

A different set of parameters appear optimum for speech; e.g., 5 milliseconds delay and the intensity of the delayed signal 3db less than the direct signal. A tested compromise of 10 milliseconds time delay and 3db attenuation yields very good over-all results. The 8- to 10db channel separation due to Precedence Effect added to the 3db intensity difference gives results comparable to a system with 12db channel separation.

It might be of interest here to mention some desirable side effects of these circuits. The literature is full of references to the "hole in the middle" and to the subjective reaction to the reproduction of the music of a full orchestra from a box occupying some 2 or 3 cubic feet.

The employment of the Precedence Effect for channel separation causes the area of the apparent sound sources to seem quite large. The sound no longer emanates from the small speaker cabinets but appears to be produced by large area sound sources. The apparent area of these sources greatly diminishes the "hole in the middle" effect and is more suggestive of the appropriate size of loudspeakers required to reproduce orchestral music.

This development should make it possible for many more broadcasters to offer double or triple channel stereo programming without diluting the stereo effect or penalizing the single channel listener, who will make up the majority of their audience.
.