CAPÍTULO 8 -- GRAVAÇÃO: A SUPOSTA 'PRESERVAÇÃO' DO SOM, COMPUTADORES NA MÚSICA, E MIDI
It is a total impossibility to "store" sound. As soon as the air vibrations die away, they are gone forever.
No, I am not ignoring the mountains of records, tapes, and compact discs all around us. When we 'record' in any medium, we do not preserve the original sound. A record, tape or compact disc doesn't contain the original sound any more than a photographer brings home the Grand Canyon in his Pentax. What IS preserved is a set of parameters that allow some particular device (like a stereo) to reconstruct an approximation of the original air disturbance.
Recall that sound consists of minute pressure changes in a relatively stable air pressure. Thomas Edison developed a device back in the late 1800's that harnessed the small but precise mechanical power of air pressure waves to 'record' their fluctuations. It did this by causing an analogous change to occur in another medium.
Translating that last statement into more common English, a stylus pushed by the air vibrations would scratch the smooth surface of a moving wax cylinder, creating little hills and valleys in the grooves. If the sound was high pitched, the stylus (needle) made very rapid scratches. If it was a low pitch, the scratches were at a much more leisurely pace. If the sound increased in loudness, the stylus' oscillations became larger in amplitude. The playback process was a simple reversal of the recording setup with a stylus being bounced up and down by the grooves in the cylinder or disk, causing a cone-shaped object to vibrate the air in patterns similar to the original sound. This wax analogy of the air vibrations came to be called the analog recording process.
When we talk about analog equipment or analog values in any medium, that means it an infinite set of values on its scale. For example: an analog watch with a sweep second hand literally has an infinite number of points that it covers between the times of 1:01 and 1:02. A digital watch will tick off the seconds--and that will be the smallest step available for that watch. For practical purposes, most people don't necessarily need to tell time in increments less than one second. Another example--an older mercury thermometer (analog) will allow the mercury to stop at an infinite number of places on its scale. A thermometer with a digital readout will likely give you a readout that rounds to even degrees--even though we know that a reading of 72 could mean 71.6, 71.668, 71.93, 72.4, 72.44487, and so on. In the above two digital examples, a compromise has been made that will not give us more accuracy or detail than needed in that particular situation.
In a fashion similar to the thermometer and watch, analog recording of sound captures the smooth back and forth flow of the air pressure (which, along with our hearing is an analog scale). The information is then stored as a constantly changing parameter, matching the smooth change of air pressure.
While the accuracy of a medium that carefully traces air pressure changes is nice, it does have some very serious drawbacks. One of them is the large number of steps and copies involved in the recording process. The air pressure pushes and pulls an element in a microphone, which creates a tiny electrical voltage which also pushes and pulls. The voltage is often amplified, sent to a mixer, sent to the recording head of a tape recorder which then turns the voltage into a fluctuating magnetic field, which then changes the alignment of particles on the tape, which then. . . hopefully you begin to see the picture. Each one of these conversion steps is slightly imperfect in its translation, and I haven't given one third of the steps typically involved between the time a sound is made and a recording is heard. The information "loses a little in the translation".
Furthermore, to accommodate the standard record or tape medium, music must be compressed into a dynamic range of 55 decibels or less between the loudest and softest notes. (Again, a decibel is a geometric rate of change. 55dB to 58dB is a doubling of intensity; 58dB to 61dB represents another doubling of intensity).
Another major problem surfaces noisily. Computer terminology is quite appropriate here. Let's call the hills and valleys in a record 'data'. When a stylus is 'reading' the data from a record groove, it picks up the information by being bounced up and down, sending a corresponding (analogous) electrical signal to the amplifier. Dust and scratches on the surface of the record are not really any different from the hills and valleys of the groove and are gleefully read by the bouncing stylus, producing a pop, click, or other obnoxious noise. While a tape won't usually develop surface noise, they will often develop "dropouts", places where the sound information gets removed. They are also prone to flutter and wobble when the tape gets slightly worn or the playback equipment gets out of alignment. These can be nearly as annoying as the noises.
One doesn't need complicated scientific equipment to hear the problems in analog copying. Most of us have tried to copy a tape. The copy never sounds as good as the original. If we try to make a copy of a copy, the sound deteriorates even more.
While there are some hefty disadvantages in the analog recording medium, his is not to say that it is a lousy process. It has served us quite well over its first one hundred years, and at its best it IS quite good and will probably still be improved, but you can see there are some flaws.
For decades, this process was well known, but there were some severe drawbacks--mostly having to do with the staggering costs of storing these numbers and the computer equipment necessary to handle data at such a rate. For example, we have to measure the air pressure at two times the rate of the highest frequency we need to preserve. As we can hear up to 20,000 cps, that means the air pressure needs to be measured over 40,000 times per second. At that rate, a standard 5.25" floppy disk could hold less than 5 seconds of music. Considering the price of the available technology, it would be cheaper to hire an orchestra to play in your living room.
In the late 1970s, the power of computers had skyrocketed and and their prices had plummeted to the point where digital recording became commercially feasible. The recording process became greatly improved because throughout the many mixing and rerecording steps, the numbers remained as much the same as the above example of copying computer software.
Even with digital recording, the weakest link in the process still remained the records and tapes used in the playback units (remember they are analog). The compact disc came along shortly afterward, keeping those numbers intact until the very last minute when they had to be reconverted to analog signals that drove the speakers. Finally, the process has reached a point to where, for all practical purposes, the listener at home is listening to an exact copy of the studio master.
The digital recording process measures the air pressure changes (called sampling). The recording industry standard is a rate of 44,000 samplings per second. Again, the process is satisfactory when the sampling rate is twice the highest frequency needed. As we can only hear up to 20,000 Hz or less (and many stereo systems can handle less than that) 44,000 has been adopted as the industry standard, balancing off several needs. While it isn't perfect, it is sufficient for the human ear. To sample every single part of the air pressure would require an infinitely large sampling rate and an infinitely large storage medium. Like the above examples of the watch and the thermometer, for most practical purposes, smaller increments would be a waste of space, technology, and probably cost.
These 44,000 numbers, (each one a 16 bit binary number, such as 1011010010010011) are then stored on some kind of tape as distinct numbers, unlike analog recording which tries to record a copy of the air disturbance. Any mixing, editing, rerecording is done keeping the numbers intact.
At some point, the numbers are reconverted (using a digital-to-analog converter) back into electrical waves that drive the speaker, causing it to push and pull against the air, creating a very good reproduction of the original sound waves. Although this part of the process is really an analog signal, it is necessary, because essentially we hear in an analog process. The recordings are much much clearer because many analog steps have been skipped.
The dynamic range that a digital recording can handle is approximately 95dB compared to 55 dB for the best record or tape. If you remember the information given about the decibel scale, this means that a CD can handle over 1,000 times the dynamic range that the best records can.
Musicians have been finding the computer valuable in a couple of other areas related to sound production. One of them has been in the enhancement of older recordings.
One of the maddeningly fascinating sets of recordings has been the set of recordings made by Enrico Caruso between 1906-1921. Caruso was one of the most powerful operatic singers of this century, if not all time. He lived long enough to make a large number of recordings, which were unfortunately done without electricity involved in the recording process. They are very noisy and the sounds are badly distorted, but give a glimpse of what this marvelous singer must have sounded like in the flesh.
In the early 1970s, an RCA recording engineer by the name of Thomas Stockham digitally recorded these old Caruso recordings. On the digital tape, as you might recall, the measurements of sound are stored in a series of numbers. And, numbers are the lingua franca of computers.
Stockham and his associates developed software that would be able to take a look at these patterns of numbers, determine which ones were noise (pops, clicks, scratches, etc.) and reduce or eliminate those sounds. The program would examine the patterns of the sounds, decide if there was any distortion, and attempt to remove the distortion, and enhance the sounds of the existing music. Keep in mind that this is not the same as using tone controls or filters--this is a complete reconstruction of the sounds on the recording. A task such as this that required the billions of calculations necessary could only happen in the age of the computer. Like the process in playing a CD, these new sets of numbers were then reconverted back to sound.
While the resulting recordings were far from perfect, they showed a definite enhancement and became easier to listen to. The process has been used on other "vintage" recordings and shows promises of improvement.
Preceding Stockham's restoration, another version of the process was used to create some interesting compositions. "The Story of Our Lives" by Charles Dodge is a example of digitized sound that has been manipulated in a highly creative and often humorous manner. Instead of sound being corrected, the recording of Dodge's voice is transformed into a half-human, half-machine oral interpretation of the text.
Another computer process (with a little murkier plan in mind) was the use of the computer to generate sound. Before the advent of synthesizers, the computer showed great promise in the generation of synthetic sounds. Like the above processes, it was wrapped around numbers--millions of them.
For the most part, musical sounds are quite predictable in their regular patterns. There is a very close correlation between pitch, amplitude, and tone color changes and the physical characteristics of the sound produced.
Composers began using programs that created huge series of numbers that mimicked the patterns of musical sounds. These numbers were stored on tape (just as in the digital recording process mentioned above). The playback process would be just the same as in the earlier example. The only difference would be that these patterns of numbers had never before been sound. This was literally music created out of thin air. Composers found it thrilling to be faced with the possibility of total control of the end product.
Insiders suggest that the whole project was for the purpose of eventually making acoustic and human musicians obsolete. No more orchestras to deal with! No more musicians' unions limiting rehearsal time, no more imperfect and temperamental prima donna conductors!
Fortunately, the project became someplace between a failure and a mere curiosity. Music and musical sound is far more complex than can be dealt with by machines, no matter how fast or powerful the computer--at least at the time of this writing, and probably for the foreseeable future.
To be sure, the project did produce some fascinating experiments with a rhythmic precision never before heard, but the electronic sounds produced were difficult for the average person to listen to for an extended time. A million or so years of evolution seems to have drilled a curious need into us--a need for the richly complex sounds of our environment--a need that is quite well fulfilled by many of our machines used to vibrate the air--in other words, our very own acoustical instruments.
Many musical experiments have left a shadow on the standard computer's ability to produce music. Most personal computers (the ones that typical musicians would have access to) have sound generation abilities that are laughable at best. As musical instruments, they lack any serious abilities.
However, an important idea developed in the 1970s--instead of using the computer as a musical instrument, what would happen if it was used as the conductor?
The result of this is MIDI: Musical Instrument Digital Interface. "Interface" is a term meaning a device that will translate signals or act as a go-between for two otherwise incompatible modules. "Digital" has to do with numbers, and "Musical Instrument" refers to exactly what it sounds like. The result is a set of hardware and software that will link musical instruments to computers, allowing capabilities far beyond either's individual abilities. MIDI has already begun to revolutionize the music industry the way that the word processor has changed the manipulation of text.
Most synths and keyboards manufactured today will have this special MIDI hookup that will allow two-way communication between them. Typically, a musician will play music on the synthesizer and the computer will store the sequence of notes played (along with other information discussed below).
The musician will then make any necessary changes to the music--like word processing, mistakes can be corrected, effects can be added, etc. This data can then be stored on disk like any other type of computer file. The musician can then tell the computer to send the information back to the synthesizer, which then "plays" the music sent to it, in essence becoming an electronic player piano. Note that in this setup, the synthesizer is producing the actual sound, not the computer. Instead of little beeps coming from a tinny PC speaker, the sound is as good and rich as the synthesizer can produce. The music can also be played back at a different tempo from when it was recorded, in a different key, etc.
At this time there are four common functions that can be performed by a MIDI system. (A) sequencing: meaning that the computer stores the sequence of notes played by the performer. This sequence includes the key velocities (loudness), and the timings between the notes and rests. It an be saved on disk, reloaded, or transmitted by modem; (B) editing: mistakes or other inaccuracies can be changed by the computer keyboard or a mouse. The composition can be transposed, tempo changed, etc. Although it is very time consuming, an entire composition can be entered from the computer keyboard. There are a number of quadriplegic composers who create entire compositions using a modified keyboard; (C) reprogramming the synthesizer: most synthesizers can receive information from the computer and change sounds or effects in the middle of a program; and (D) notating: with certain programs and high quality printers, it is possible to print out a page or a full score of music with a quality that rivals that of a commercial music publisher.
Another important, but less common, use of MIDI is to allow the computer to "create" a little bit based on the performer's music. A number of clever pieces of software have been written that will analyze a person's improvisations and begin mimicking his style. However, musicians need not feel threatened by this. So much of music is based around what we sense and perceive in our environment that there is little danger a computer will be able to latch onto these things and be able to handle the variety of a breathing musician.
The real talent of MIDI lies in the fact that standard systems can send information out on sixteen different channels--that means that the musician can record sixteen separate parts, and if he has sixteen different synthesizers or drum machines, etc., he can set them to receive only one specific part and play a specific line.
If you think about that a little, yes--that gives one person the power to play every instrument in a rock band, a small jazz ensemble--all the way to being able to cover the complexity of a small to medium symphony orchestra.
As the computer and the word processor have given us all the power of the publisher, will MIDI give everyone the power of the conductor? What will happen when people who are barely literate in music can create thick, complex, impressive sounding compositions using high-tech equipment?
Those are some questions that legitimately worry a lot of musicians--unnecessarily.
A decade after word processors, spellcheckers, grammar checkers, style checkers, and laser printers, everyone who can read and type should be able to become an author.
It simply hasn't happened and it's not going to happen.
What this software and hardware has become is a tool that allows people to extend their creative reach and make better use of their time--in theory, because using computers doesn't necessarily guarantee these things will happen. However, if a person doesn't have that creative spark, or at least a little ability, nothing will come from the most powerful inventions. Those who have the talent, skill, and diligence to write will continue to do so, perhaps at a faster pace and with a few less misspelled words. For the rest, it will make little difference. Efficient use of a computer is yet one more skill that will need to be mastered. Regardless of the ease in using the software, this is something that many will always find troublesome.
There is absolutely no reason to believe that anything different will happen with MIDI.
To create a truly "musical" performance requires an understanding of the score far beyond simple performances of the printed notes. There are far more things to a musical performance, or even the creation of music than just slapping together a few notes or harmonies. In theory, MIDI would allow the average near-non-musician to perform a Beethoven symphony, but few would tackle such a Herculean task. The end result would hardly be satisfying. As in the example of word processing versus writing, MIDI is acting as a tool that will help its users sharpen and refine their talents, perhaps making things faster for a few--the few who would probably have done it anyway.
Besides, if a non-musician did tackle such a project, one of two things might happen: he might produce a very mechanical, coldly precise performance--in other words, lacking in everything we consider to be "musical"; or along the way he would probably learn so much as to eliminate many of the differences between himself and a musician--which wouldn't be a bad thing.
MIDI does open up some exciting possibilities. Musicians do not need to be on the spot to create or perform music. A personal example--a friend (with a MIDI system) needed some music recorded for a video he was producing. He called me, told me the requirements, and I got to work. When I was finished, I saved my work to disk, called him back, and transmitted the file over the phone by modem. A few minutes later, he called me once more--and the music I had just created was playing in the background. Jan Hammer worked the music for the TV series Miami Vice in a similar fashion, not being within 1000 miles of the production.
If nothing else, one thing is certain--MIDI is going to create a generation of composer/performers who are equally literate in music and computer skills--which is not such a bad idea, anyway. . .