|
|
|
|
Tech Topics Technical Discussions of Audio and Loudspeaker Issues |
Visit Our Audio Engineering
Book Store |
An Overview
of Echo, Reverberation and Other Delay Effects |
|||
|
|||
| For
the past few months we've been reviewing various audio products which generate delay
effects and/or reverberation; and we will be reporting on several more of these units in
subsequent columns. This month, however, we would like to pause and review some of the
basic audio and acoustics concepts important to understanding the operation and effective
application of this equipment. This article should be helpful to those considering the
purchase of such equipment, especially considering the variety of devices available. We'll start by discussing echo, then reverberation and, finally, other delay effects. In each case a definition of the effect will be given followed by discussions of: how the effect is produced, applications of the effect and audio quality evaluation of the effects. ECHO An echo is simply a slightly delayed repetition of a sound. In our acoustic environment echoes are heard as the result of sound waves being reflected from a surface and returned to the listener. The time difference between when the listener hears the direct sound and when the reflection (echo) is heard is referred to as the echo delay time. This delay time depends on the distances between the listener, the sound source and the reflecting surface, with greater distances resulting in longer delay times. For example, consider a listener outdoors standing 100 feet from a large reflecting surface (the side of a building maybe). The listener snaps his fingers and hears the snap followed immediately by a single echo. The echo delay time can be easily calculated by considering that sound travels in air at a speed of about one foot per millisecond (msec., one-one thousandth of a second). It therefore takes the sound about 100 msec. to travel the 100 feet to the reflecting surface and 100 msec. more for the reflection to return to the listener. Therefore, the echo delay time in this example would be about 200 msec. |
This delay is long enough for the echo "snap" to
be heard as separate from the initial, or direct "snap." That is, the listener
perceives two distinct snaps. As the echo delay time is shortened, a point is reached
where the listener no longer hears two "snaps" but instead the two sounds fuse
together and are perceived as a single sound. It has been
observed that this occurs for delay times shorter than about 30 msec. The exact delay time
where a distinct repetition is no longer heard is somewhat dependent on the nature of the
sound and is shortest for short duration sounds- "clicks" or "pops,"
for example. Echo is a sound phenomenon that we are most accustomed to in our outdoor environment; whereas indoors, the sound reflections are so numerous that we usually hear "reverberation" rather than many distinct echoes. It is the inaudibility of the distinct repetitions that distinguishes reverberation from multiple echoes. This can be easily understood in terms of the 30 msec. rule. In a room, the sound reflections arriving at a listener's ear are very closely spaced in time, much closer than 30 msec., and are therefore not heard individually. Instead, the many closely spaced reflections create a different effect: "reverberation" (more on reverb later). The first popular technique for artificially creating echo made use of the tape recorder. The sound signal was recorded on tape and then immediately played back to provide a copy of the original signal delayed by the amount of time it took the tape to travel from the record head of the recorder to the playback head. When the delayed signal was mixed with the original signal and auditioned over a loudspeaker, the delayed signal was heard as an echo of the original signal. It was quickly discovered that if a portion of the playback signal was fed back to the record head then a series of echoes would result and the resulting effect was found to be musically useful. If either the tape speed or the spacing between the record and play heads was made variable, then the echo delay time could be adjusted for the desired effect. Likewise, the portion of the playback signal fed back to the record head was |
||
page 90
| usually made variable so the
user could control the echo repeat characteristic. Over the years since its introduction
tape echo has continued to be a popular musical effect, both for recording and for
performing. The decade of the seventies brought with it integrated circuit technology. Out of this advanced electronics emerged two new techniques for providing time delay of audio signals by purely electronic means: analog delay and digital delay. Both techniques begin by "sampling" the audio signal at very frequent time intervals (about 50,000 samples per second for good audio quality) and then perform delay processing on the individual samples. Although the techniques for performing "sampled delay" were available well before the seventies, without integrated circuit technology the sheer quantity of discrete devices required would have been prohibitive. In the analog delay technique, the samples are delayed by shifting the signal sample through hundreds (or even thousands) of sample holding stages at a high rate of speed. The total number of stages and the rate at which the sample is handed from one stage to the next determine the delay time of the system. As the samples emerge at the end of the delay line they are combined to recreate the input audio signal. The real beauty of this technique lies in the fact that these many hundreds of delay stages are all contained on a tiny little silicon chip (integrated circuit, or simply "IC") that you can buy at your local electronics shop for about $10. Welcome to the space age! The digital delay technique is similar to that just described for analog delay except that the samples are first converted to binary numbers (i.e., numbers made up of "ones" and "zeroes" like: 01101011) or "digitized" before entering the delay line. So, rather than passing a continuously variable voltage from stage to |
stage as in analog delay, the
digital delay passes a string of numbers ("digits") from one stage to the next.
At the end of the line these numbers are individually converted back to voltages and
combined to form a time-delayed replica of the input signal. Don't forget, this is all
happening at a rate of about 50,000 samples per second. Pretty clever, wouldn't you say? As with tape echo, both the analog and digital delay lines can be fed back, or "recirculated" as it's frequently called, to produce multiple repeats that decay in intensity at a rate dependent on the amount of recirculation used. A typical echo unit as might be seen on the market today would provide front panel controls for: signal input level, echo delay time, echo recirculation and mix of direct signal with echo signal. The latter control allows the user to pan between the direct sound and the echo output, thereby providing any desired mixture of direct sound with echo effects, ranging from direct sound only (no effects) to effects only (no direct sound). In addition to these controls, various available units include signal level monitoring ranging from a simple peak overload indicator to a wide-range LED level display, along with output level controls, selectable instrument or line level inputs and options for use with either balanced or unbalanced signal inputs and outputs. Although there are still tape echo units available, the analog and digital delay types are rapidly making them obsolete. The audio quality of an echo unit can be evaluated in a straightforward fashion. The delayed signal should simply be an exact copy of the input signal with respect to: distortion, noise, frequency response, bandwidth, wow and flutter (for tape units) and slewing headroom. In the case of tape delay units the audio quality of the echo is simply the same as that for a signal passing through the record/play chain once. When recirculation is used, however, the audio quality is degraded for each repeat to the extent that it would be degraded by several tape generators. Distortion and noise increase and bandwidth decreases. The same is true for analog and digital delays, but digital delay techniques are capable of a higher level of audio quality than typical tape record/play systems. With higher quality for one pass through the delay line, on successive passes (recirculation) the quality is not degraded as quickly. Now, this doesn't mean that digital delay units always provide higher quality than |
page 91
| tape or analog delays. In fact,
both analog and digital delay units (with a few notable exceptions) frequently compromise
high frequency bandwidth for economy and delay time. In contrast, tape delay can provide
full high-frequency bandwidth in conjunction with long delay times. The one outstanding
advantage that the analog and digital delays have over tape is that they utilize nothing
but solid state electronics. They have no moving parts, require no routine maintenance and
can be highly reliable. Tape machines, on the other hand, utilize intricate
electromechanical systems with many moving parts (and attendant wear) and require that
their heads be cleaned routinely and the tape periodically replaced. REVERBERATION Associated with interior spaces (usually rooms) is the natural acoustic phenomenon we call "reverberation. " In contrast to "echo," where a sound is repeated and the repeat is heard as distinct from the direct sound, reverberation consists of very many repetitions of the original sound spaced so closely in time that no single repetition is distinctly audible. Rather than repeat the original sound, reverberation (or simply "reverb") has the subjective effect of making the sound "finger" and die away slowly. Auditoriums and concert halls are frequently evaluated on the basis of their "reverberation time, " that is, the time that it takes for a sound to decay away to inaudibility. If the reverb time is too long, the room can make speech difficult to understand; too short a reverb time will make a room sound dry and lifeless. In the recording studio it is frequently desirable to add reverberation to recorded tracks. This can be done by using either an acoustic reverberation chamber or by employing some sort of artificial reverberation unit. The acoustic reverb chamber consists of an isolated room equipped with a loudspeaker and a microphone. The recorded track to which reverberation is to be added is played back through the loudspeaker, thus stimulating the chamber's natural reverberation. The microphone is used to pick up the reverberation in the chamber and is normally placed away from the loudspeaker to minimize the pick up of direct sound. The signal from the microphone is then amplified and returned to the control room where it is mixed with the playback signal so as to add just the desired amount of reverberation. Acoustic reverberation chambers are usually used only where the highest quality reverberation is needed because they require the dedicated use of a fairly large room. There are two basic families of artificial reverberation devices: the electromechanical systems (spring and plate type reverbs, for example) and the newer discrete time delay devices (ranging from simple delay line with feedback to fairly complex digital reverberation systems). At this time there are probably many more spring reverbs in use than any other type. This is because of |
their low cost and small size in
comparison to other units. But the small inexpensive spring reverbs also provide the
poorest quality reverb; whereas the larger, more expensive spring reverbs can provide
reverb quality approaching that of good concert halls. The operation of the spring units is quite simple. A spring is suspended on an isolated mount and is excited at one end by a loudspeaker-like transducer which transforms the audio signal into vibrations in the spring. The sound waves induced in the spring then reflect all along the length of the spring and excite many reflections, much as sound in a room excites many reflections. Finally, the sound waves in the spring are detected by a microphone-like transducer at the other end of the spring; then the reverb signal is amplified and made available at the output of the unit. The signal is returned to the control room and mixed with the direct signal in the desired proportion. The plate reverb seems to be the unit of choice in many recording studios because of its bright, highly diffuse sound. Unlike some spring units, plate reverbs are neither inexpensive nor small; but they provide consistently high-quality reverberation. Don't get the wrong idea about spring reverbs though, at the high cost end of the spectrum are some truly excellent spring reverberation units, clearly competitive with plate units in the sound quality of the reverberation they provide. As with spring units, the plate operates by being excited by a loudspeaker-like transducer at one end setting up sound vibrations in a large steel plate which is carefully suspended in an isolating enclosure. The sound waves are reflected throughout the plate much as they would be in a highly reverberant room. The reverberant sound field is then detected through the use of a microphone-like element attached to the plate. The reverberation signal is then amplified and provided to the output of the unit for return to the control room. For best results the plate enclosure should be located in an isolated environment to minimize the pickup of stray sounds. Another type of electromechanical reverb is the "foil" reverb which is similar to the plate reverb except that the steel plate is replaced by a sheet of gold foil. The second family of reverberation devices, the discrete time delay devices, is quite young. The first of these appeared shortly after the first digital delay units hit the market. In fact, many digital (and analog) delay lines have provisions for feedback and indeed claim to provide "reverberation" when used with large amounts of feedback. This can be misleading, since the output of a simple recirculated delay line typically sounds very "electronic" and not at all like a room. But there is at least one digital reverb currently available which uses a highly complex system of recirculated discrete time delays to provide very high quality artificial reverberation. It's difficult to make an objective evaluation of the
|
page 92
| audio quality of a reverb
because so much has to be based on subjective listening tests. An overall impression of
the frequency response of the unit can be obtained by driving the unit with pink noise (a
test signal containing random noise with equal energy per octave bandwidth) and observing
the energy distribution of the output reverb signal. For minimum coloration the output
should have a flat amplitude response over the audio spectrum. This only tells part of the
story though. It's also important to know how the frequency balance tends to change as the
reverberation decays away. If the lower frequencies decay more slowly than the high
frequencies, then the reverberant sound will have more bass as it decays and may tend to
sound "warm." On the other hand, if the highs decay more slowly, the sound will
have more treble as it decays and may tend to sound "bright. " An idea of the
character the reverb will take on as it decays can be obtained by measuring what's called
the "signature" of the reverb, or the decay time of the reverb in one-third
octave bands. Those frequency bands with longer decay times will tend to dominate as the
sound decays. Again, for minimum coloration we would want the reverb signature curve to be
flat over the audio spectrum. About the only other meaningful measurement we can make on a
reverb unit is to measure the noise level at its output compared to its nominal signal
output level. There are only a few front panel controls on most reverbs. It's appropriate to provide an input signal level control and some sort of signal level indicator. In addition, some of the better spring units, and most of the digital units, provide a means of adjusting the decay time. The best units provide some degree of control over the reverberation signature, typically allowing independent variation of the high and low frequency decay times. This provides a great deal of control over the coloration of the reverb. Some units also provide multiple band equalization on the reverb output; this allows the user to adjust the overall frequency balance of the reverb but does not affect the way the reverb decays (i.e., the reverb signature). Artificial reverberation devices have become so widely used and accepted that a recording studio is hardly considered complete without one (or more). It's only necessary to look at the prices of the best reverbs (from a couple of thousand to several thousand dollars) to realize that the ear is highly sensitive to reverb colorations, and that the industry is willing to pay big bucks for truly excellent sounding reverberation. OTHER TIME DELAY EFFECTS Echo and reverberation are the best known and most widely used time delay effects, probably because both effects occur naturally in our acoustic environment. Besides these two effects, there are other musically useful time delay effects which are not prominent in our environment but which, none the less, can be artificially created through the use of short time |
delays. Among these effects are:
"flanging," "doubling, " and "chorusing. " The effect popularly known as "flanging" was first produced with the aid of two tape machines. A signal was fed to both tape machines where it was recorded while monitoring off the tape continuously; the outputs of the two tape machines were then summed and the summed signal auditioned over a monitor speaker. With the two tape machines running normally, there would be no effect on the sound. However, if one machine was made to run slower than the other (by dragging a thumb on the flange of one tape supply reel) it was observed that a very interesting, and musically useful, "swishing" sound appeared as an effect on the original sound. Analog and digital delay lines have now made it unnecessary to use tape machines to do flanging, and as a result flanging has become a very popular effect both for recording and for on-stage use. Flanging can be explained as follows: When two identical signals are combined there are normally no unusual effects. However, if one signal is delayed slightly in time and then the [two] signals are combined, the net result is a series of cancellations of certain frequencies, the canceled frequencies depending on the precise amount of time delay between the two signals. This series of cancellations is frequently referred to as a "comb filter" because the frequency response plot of such a filter has many sharp notches which resemble the teeth of a comb. The most interesting part of the effect occurs when the delay of the one signal path is made to change with time. This causes the comb filter to sweep through the audio spectrum and imparts the unique "swishing" sound associated with flanging. When the time delay is increasing the notches sweep from high to low, and as the delay is decreased the notches sweep low to high. The time delays required for the effect are quite short, about one millisecond. However, sweeping the comb filter through the spectrum requires that the delay be varied from about 0.1 msec. up to about 10 msec. The flangers available on the market automatically vary the delay time so that the comb filter automatically sweeps up and down through the spectrum. The user is typically provided control over both the range and the rate of the sweep. The user is also provided with a pan control which allows the flanger output to be varied from all direct signal to all delayed signal or anywhere in-between. The maximum flanging effect is obtained with an equal mix of the two signals. Most flangers also provide for recirculation of the delayed signal which provides an interesting variation on the effect. "Doubling" is an electronic delay effect which attempts to copy the sound of the recording studio trick known as "double tracking." Double tracking is done in the studio by separately recording two identical musical parts and then playing back the two parts together. For example, a vocalist's lead line is recorded on track one of the recorder; then the tape is rewound and the vocalist gives a second performance |
page 94
| [of the same part] which is
recorded on track two. When tracks one and two are played back together the impression
might be that the vocalist and his identical twin [clone7] are singing together in unison,
but more often the effect is perceived simply as an enrichening or "fattening"
of the original track. Doubling is an attempt to copy the double track effect without all
the hassle of recording two versions of a performance. Also, note that double tracking is
an effect that can only be used in recording, that is, it can't be used in a
"live" performance (unless the performer has an equally talented twin who
doesn't mind hiding). The currently available devices for doing doubling basically add a short delay (10-20 msec.) to the direct signal. This does not, by itself, constitute a very convincing doubling effect because the "second track" (the delayed signal) is a perfect copy of the original-too perfect in fact. In the studio, when double tracking is done the second performance is always slightly different from the first; so if we're to duplicate the sound of double tracking we need to find a way to make the delayed signal somehow different from the direct signal. The most popular way of "perturbing" the delayed signal is to slowly vary the time delay over a small range. This has the effect of alternately driving the pitch of the delayed signal sharp and then flat by a slight amount thereby introducing a small musical |
difference
between the direct and the delayed signal. With this improvement, electronic doubling can
be fairly convincing, but none of the units we've heard can provide quite the same effect
as that obtained with studio double tracking. The effect generally referred to as "chorusing" is an attempt at producing the impression of many voices singing or many instruments playing in unison, somewhat as if a performance had been double tracked many times over. Chorusing is provided through delay devices by using the approach described for doubling and then recirculating the delayed signal to provide multiple repeats. The results seem to be highly variable in effectiveness. CONCLUSION We hope this overview of time delay effects will make future reviews of echo and reverb equipment more meaningful to our readers. In particular, we hope to have made clear the distinction between echo and reverberation and the types of equipment used to produce these effects. Let us note here that the delay effects units typically used to produce echo, flanging and doubling, do not, in general, provide high quality reverberation and should not be purchased solely for that purpose. They do, however, produce some exciting and very useful effects. |
|
Reproduced from Modern Recording & Music magazine, May 1980.
|
|
|
|
|
True
Audio Home Page |
Catalog |
Tech
Topics | Audio Links |
Book
Store
Your comments are welcome at
webmaster@trueaudio.com |
