THE ART OF RECORDING THE BIG BAND
by Robert Auld
(This paper is in two parts. The notes for each part are included on the same page as that part. Notes are indicated by a number in parenthesis.)
What is the jazz big band?
A history of recording the big band: the 1920's
The 1930's: "swing" and electrical recording
An example from the period: Benny Goodman at Carnegie Hall, 1938
The 1940's: a decade of changes
The long road of the LP
Mullin kidnaps the Magnetophon
The LP arrives
The 1950's: the marketing of High Fidelity
The engineering consensus
Stereo: early efforts
The big band in stereo: a different approach
Three tracks on tape
The engineering style of the three-track era: a few examples
The temptations of complexity
Notes for Part 1
My interest in the big band dates from childhood. My father was a Benny Goodman fan, so I grew up with his music in the house. I became a musician myself, and so I got to play in big bands. I was interested in audio recording, and I tried recording big bands. I was not very satisfied with the results. So, I paid attention to how others were doing it. I have been paying attention for about 25 years. Some of what I have learned is set down here.
Part 1 of this paper traces the history of big band recording, from the 1920's up to the mid-1960's. Part 2 deals with the changes wrought by the introduction of multi-track recording methods in the mid-1960's, the problems those changes caused for big band recordings, and some suggested remedies. Finally, Part 2 closes with the analysis of several modern big band recordings.
Common variations have included reductions (Benny Goodman initially used only 3 trumpets and 2 trombones for a five-piece brass section), or various doubles and additions, such as:
Rhythm: vibraphone; various latin percussion including congas, bongos, etc.; even tympani on occasion. Since the 1970's, bass guitar has often substituted for the upright bass.
Reeds: it is very common to have the saxes double on soprano sax, clarinet and flute. Bass clarinet, oboe and bassoon have been used occasionally.
Brass: flugelhorn is a common double for the trumpets. Tuba is sometimes added to the trombones. From 1 to 4 french horns have been added. In multiples they are usually treated as a separate section.Big bands have also included string sections, vocal groups, accordions, rock guitars, synthesizers--in short, just about anything you could possibly imagine. My emphasis will be on the ensemble as it is classically understood, with glances at various additions, as needed.
Acoustic recording could not reproduce the full frequency range of instrumental tone. Therefore, it was common to alter the instrumentation of ensembles to suit. For example: tuba was usually used in place of string bass, and drummers would discard their bass drums and use only higher pitched instruments like wood blocks and cymbals.
When electrical recording was introduced in 1925, these practices continued for awhile. Due to a lack of good playback equipment, many engineers did not realize at first that the new technique could actually reproduce a much greater range of tone. Fortunately, this changed fairly quickly.(1) From about 1928 on, instrumental groups recorded with the same configuration as they used in live performance. Ensembles were balanced by grouping the musicians at various distances from the single recording microphone. A musician taking a solo would walk up to the mic and play into it, then move away for ensemble passages.
Instead of "big bands" it would be more appropriate to talk about dance orchestras. The larger ones were evolving into what would become the big band of the 1930's. (2)
Electrical recording was still at an early stage of development. The great range of equipment that we are accustomed to today did not exist then. Studios independent of record companies or radio stations were rare. In any case, studio electronics were often built in-house, and elements that could not be easily built (microphones, disk cutting lathes, etc.) were obtained from a few large suppliers such as RCA or Western Electric. Much of the equipment was developed with the needs of the film and radio industries in mind, but was readily adapted to the needs of record studios. (In fact, recording sessions were sometimes done in radio studios, as most of the usual equipment was already at hand.)
Remote recordings were usually associated with radio broadcasts. Again, the same standard-type equipment was used. Therefore, the signal chain for almost any late 1930's big band recording might be as follows:
-One or two microphones, typically either condenser element, omni-directional pattern (Western Electric made a number of models) or ribbon element, bi-directional pattern (RCA type 44, for example).
-Console, custom-built, or a modified broadcast board. Four microphone inputs would be about as extravagant as the designers would get, in most cases. Mono output.
-Disk cutting lathes, either 78 rpm or transcription format. Commercial 78 format was about a 2.7 mil. groove cut laterally, disk size up to 12 inches. Professional transcription disks, as developed by Bell Labs in the mid-1930's, were vertical cut ("hill and dale") with diameter up to 16 inches and could be recorded at slower speeds for more playing time. The largest transcription disks could hold almost 30 minutes of material and were commonly used for delayed broadcast of radio shows.
-Signal processing (equalization, dynamics, reverb) was initially available only in very primitive form. For example: electrical disc cutting used pre-emphasis to reduce surface noise and allow greater playing time. In some cases, the engineer would alter the crossover point and/or slope of the recording emphasis of a given disk if he felt he would get a better result that way. Therefore, playback curves for electrical 78's tended to be nominal (which poses problems for restoration of such recordings).
-Dynamics were controlled by manual gain-riding, moving the microphone, instructing the musicians to play louder or softer. Compressors and limiters did not become widely available until after World War II.
-Reverb was usually a matter of the room one was recording in. At least one film studio had spring-type reverb units custom built in the mid-1930's, but outside of Hollywood or Bell Laboratories such experimental equipment was hardly even thought of. Radio serials, with their need to portray many different locales, inspired the first live echo chambers (often, converted bathrooms or stairwells). Such enhancement was not usually available or even considered desirable for straight music recording until the 1940's.Despite the limited equipment available, the better records of the period give a pretty clear picture of how the swing-era big bands sounded. The best engineers of the time had an impressive feel for what was important in the music and were able to capture the essentials. And the simplicity of the methods they used insured a certain honesty in the recordings. With no overdubbing, no splicing, and very little in the way of tonal enhancement, what the musicians played is what you get.
From the technical point of view, the 1938 Carnegie Hall Concert is typical of the remote airchecks of the time. Two microphones were used, one hanging over the stage for overall pickup, the second on stage and also used for the P.A. system. These two sources were probably mixed on the site, and the signal then sent to CBS studios (using the usual broadcast-remote transmission lines) where the transcription discs were cut. The setup differed from a radio remote in that there was no announcer, and therefore no separate announce microphone.
Besides its historical and musical value, this recording teaches us about something else: the live acoustical balance of Benny Goodman's band as compared to the balance heard on their studio recordings. The brass and drums really tend to dominate the mix. The relatively distant placement of the overhead microphone lets us hear this clearly and confirms the critical consensus about Goodman's band: it really was louder and brassier than what had come before; a fact that caused Goodman problems with some hotel ballroom managements early in his career. (3) Also evident is the wide dynamic range of the music, which often threatens to outrun the limits of the recording equipment.
In the studio the engineers frequently placed the softer instruments much closer to the recording microphone, the result being an apparent reduction in dynamic range. But in this live concert the true capabilities of the classic big band were on display. It was (and is) a formidable ensemble capable of extremes of dynamics, which arrangers readily exploited. The recording methods of the 1930's could only give an indication of what was there, and the tendency of studio big band recordings, even now, is to reduce the range between loud and soft. (4)
The phono cartridge problem was first tackled at Harvard in the late 1930's. A team of engineers there developed high quality disk recording systems. The need to play back the resulting disks without damaging them led to the development of new cartridge designs. This work established the basis for the modern phono cartridge. (6)
The source material problem would eventually be addressed with the introduction of high quality magnetic tape recording. While there were attempts to develop tape recording in the U.S., the results were not of high quality, and were used mainly for spoken voice applications by broadcasters. (7) But in Germany, during the early 1940's, high quality machines became available, even including some experimental stereo machines. (8) Due to the on-going war, we had no knowledge of those developments. So when the tape machines were discovered at German radio stations after the war, it was something of a shock.
After he returned home from his tour of duty, he assembled the machines and arranged to demonstrate them for several recording and broadcast professionals. The first development was the recording of Bing Crosby's radio show on the Magnetophons. This experiment was so successful that Crosby's producers decided to use tape as the initial recording medium, instead of transcription disks, for all future shows. (The shows were still distributed on disk as before, but tape editing and copying were used for the preceding production steps.) The second development was the production of what was essentially a copy of the AEG Magnetophon design by an American company that specialized in the production of precision electric motors. The company was Ampex, and the new machine was their first professional recording product. (9)
The new medium had far lower noise and distortion than the old 78's. Sonic details that were considered of little importance for 78 production (such as acoustical ambience) assumed new importance in recordings made for LP issue. The immediate effect was a general upgrading of the quality of the recording chain, with more attention paid to studio acoustics, reverb devices, etc.
Microphone placement practices for big band sessions changed only incrementally in the early LP era: there would be an overall pickup mic, spot mics on the piano and the bass, a mic or two for soloists; a total of 4 to 5 microphones was typical. (10)
(Record company executives were not, at first, certain if the public would take to High Fidelity. It took psychoacoustic experiments by Harry F. Olson, VP of RCA labs, to convince them. See note (11).)
This was not new, but the degree of importance was. In the 1930's and 40's, the conditions of playback enjoyed by consumers were usually so poor that even if a given company's records were exceptional in the engineering, few were likely to notice. By the mid-1950's, the situation had changed sufficiently that the promotion of "Hi-Fi Spectaculars" was a common element of marketing recordings.
"The Benny Goodman Story" gives us the opportunity to compare the studio practices of the mid-1950's with those of the mid-1930's, using identical repertoire, played by many of the same musicians. (12) The first impression of such a comparison will tend to vindicate George T. Simon. The newer recording has a smoothness and clarity that is startling when compared to the old 78's. There is some evidence of a more complex microphone set-up: the saxophones sound separated from the brass, an impression confirmed by the engineer raising or lowering their relative volume a bit here and there. The bass and piano have a clarity usually associated with close microphones. And yet, what I find most impressive is the similarity of the sound to the older recordings: the balance between brass and reeds is as before, the drums tend to be in the background, Goodman's clarinet is solidly in the foreground.
This basic consensus would hold up pretty well until the late 1960's when there would be a great unraveling, for reasons that I shall detail later on. But now, it is time to consider where the engineering action really was in the 1950's. The consumer was not aware of it yet, but developing behind the scenes was--Stereo.
After World War II, the introduction of tape recording made the recording of stereo masters a practical possibility. The first major record company to pursue this in a big way was RCA, largely due to the efforts of John Pfeiffer. (14) The initial RCA stereo program was confined to classical orchestra recordings. Given the large amount of money spent on any orchestra session, adding a second engineering team to record the session with the new stereo equipment was a reasonable expense. And it simply makes sense to focus your best technical efforts on the flagship product. For RCA, the jazz big band was a part of popular music, and could not command the kind of attention and expenditure that went to the Red Seal classical program. The cultural shift that would change all that was many years in the future.
In 1957, when it became clear that the stereo LP would soon be introduced (the necessary disk cutters and playback cartridges were known to be under development), adapting the standard big band setup to the needs of stereo proved to be simple. The sax section (which already had its own mic or two), was placed on one side of the stereo picture, the brass were placed on the other side and the rhythm section occupied the middle. For many studios, this meant virtually no change from their monophonic setup practices. This helped ensure mono compatibility, which was very important, as the vast majority of consumers would still have mono playback systems.
Sinatra's position in the photo is obviously for the photographer; I doubt that the engineers would have had him stand in front of the brass section! But for the rest, the conservative section-by- section pickup is evident: the brass to the right in two rows, picked up by a single mic, the saxes to the left, grouped around a single RCA 44, the rhythm section instruments roughly in the middle with single mics for each instrument, and the strings some distance to the left with perhaps two mics on them.
What we hear on the record corresponds to the visual picture almost exactly. There is some audible leakage between mics, but the matching of the stereo picture to the actual placement of the musicians prevents any strange effects. Finally, the presence of some "spread" in the stereo placement of the rhythm instruments convinces me that the "middle" recorder track was reserved for Sinatra--a wise choice, given the demands he might make.
Listening to this record, it is clear that important elements of the engineering consensus survived the introduction of stereo. The balances between the brass, saxes and rhythm instruments are similar to Benny Goodman's 1955 sessions. And in one respect nothing had changed: the consumer product was still the LP, with its familiar limitations.
Two years earlier, Miles Davis and Gil Evans recorded "Porgy and Bess" in Columbia's 30th street studio in New York City. The recent Columbia/Legacy reissue (CK 65141) has a photo showing part of the setup for one of the sessions. Miles Davis is in the foreground, the trombones can be seen just beyond to his left, and the french horns are beyond the trombones, facing them at a 45 degree angle (which causes their backwards-facing horn bells to fire well away from the other brass). A single mic behind the horns picks them up, another mic picks up the trombones (and perhaps the trumpets, out of sight behind them) and Miles, of course, has his own mic.
The sound picture of the recording does not precisely match what we see in the photo, as the horns are heard to the left of the mix, as if behind the reed section. I suspect there was sufficient isolation in this setup to move the apparent image of the horns as desired. In most important respects, though, we see the same techniques as were used in the Sinatra recording: section by section pickup, with close mics only for soloists or weaker instruments.
Additional evidence of the recording techniques of the time can be found in Harry Olson's "Music Physics and Engineering".(15) In chapter eight he diagrams typical setups for several kinds of recording sessions. His fig. 8.32, "arrangement of the instruments and microphones for recording a dance band or popular music orchestra" shows a big band arranged much as they would be in live performance, picked up with 6 mics (including one for a vocal soloist) mixed to 3 channels. Olson comments about this arrangement, "Each microphone covers one instrument or a group of instruments with very intimate coupling between the microphone and the sound source. In the case of popular music a high order of definition is desirable. Because of the fast tempo of the music, the ratio of direct to reflected sound must be kept large or the reproduced music will be blurred." (16)
An important clue is provided by Olson in his text accompanying the diagram. He states, "The use of multiple microphones makes it possible to obtain any desired balance of the various sections of the orchestra." Apparently, obtaining a reasonable pickup of the balance normally provided by the orchestra and conductor was not enough; the engineer must be able to drastically change that balance on demand. Now, if the discussion were about film soundtrack recording, the need for such a capability would be obvious. Film sound serves the needs of the story and the picture, and so needs to be constantly flexible. But here we have the same idea being applied to straight orchestral recordings of the standard classical repertoire. What was going on?
2. For examples of late 1920's dance music played by large dance orchestras, I suggest the listener turn to recordings of Jean Goldkette and His Orchestra or of Paul Whiteman and His Orchestra. A good selection of both groups can be found on the CD, "Bix Beiderbecke--Bix Lives!", RCA/Bluebird 6845-2-RB.
3. In his liner notes for "The Benny Goodman Story",(12) George T. Simon recounts how, in 1935, Goodman and his band replaced Guy Lombardo at the Roosevelt Hotel in New York City, and were promptly fired for playing too loudly.
4. "Benny Goodman Live at Carnegie Hall" is available on two CD's, Columbia G2K 40244. The CD transfer is somewhat rolled off at the frequency extremes and benefits from a bit of boost in the bass and treble. For an example of the band's dynamic range, listen to "Big John's Special", track 11 on disc 2.
7. Joel Tall, the CBS engineer who invented the tape splicing block, told of working with tape recorders made by the Brush company in his article, "Tall Tales", in Audio magazine, October 1978. The recorder, the Brush BK-401, recorded on fragile paper-backed tape. The stock model had a signal-to-noise ratio of about 35 dB and lots of distortion. By removing the power transformer from the machine and attaching it by a remote cable, Tall was able to get the S/N ratio to about 55 dB. He tried applying negative feedback to lower the distortion, without much success. (I suspect this machine didnot use high frequency tape biasing.) Even so, it proved useful to CBS for editing work during the mid-1940's, and was even used to feed the radio network directly on occasion.
8. In Audio magazine, August 1993, Bert Whyte described how Helmut Kruger, a German radio technician, made stereo recordings of orchestra concerts in Berlin with a modified AEG Magnetophon, starting in 1943. Out of the hundreds of tapes he made, about five survived the war. Excerpts from these tapes were transferred to CD by Harmonia Mundi Acoustica in a limited edition, distributed to members of the Audio Engineering Society on the occasion of their 94th convention, held in Berlin.
10. Examples of such practices can be heard in Benny Goodman's recordings of that period for Capitol Records, available on "Benny Goodman: undercurrent blues", Capitol/Jazz 7243 8 32086 2 3. This CD contains most of the studio sessions of Goodman's "be-bop" band from 1947 through 1949.
I also recommend "Glenn Miller-The Lost Recordings", RCA Victor 09026-68320-2 (issued 1996). This two-CD set contains transcriptions for broadcast made by the American Band of the Allied Expeditionary Force (Miller's war-time service big band) in late 1944. The recordings were made in London in HMV's Abbey Road Studio No. 1 (still in use today for large-scale symphonic recordings). A conservative multi-mike approach was used with single microphones on brass, reeds, strings (a 20 piece string section!) and rhythm, along with an announcer/vocal mike. There is somewhat heavy-handed use of the CEDAR noise reduction system, but it is still evident that these are among the best big-band recordings Glenn Miller made, both technically and musically.
11. At first, record company executives were not certain that "high fidelity" would be desired by their customers. Tests carried out by broadcast engineers in the mid-1940's seemed to indicate that most listeners preferred playback of speech and musical materials through systems with restricted bandwidth as compared to systems with full bandwidth reproduction. If this were indeed true, there would be little point in mass-marketing high fidelity recordings and playback systems.
Harry F. Olson of RCA laboratories suspected that the studies indicating listener preference for low-fi were themselves flawed. He believed that as the playback system bandwidth was increased, the distortions inherent in the source material became more apparent and more irritating to the listeners. As the inventor of the ribbon element microphone and the designer of the professional ribbon element microphones that RCA subsequently marketed, Olson was uniquely placed to pursue such a question.
Olson's acoustic fidelity test
In 1947 he devised an ingenious all-acoustical experiment to test his thesis. He used a small room about the size of a typical domestic living room as his test site. A variable acoustic barrier was built across the room. This barrier contained vanes that could be opened or closed. In the open position, there was no restriction on transmission of sound to the other side of the room. In the closed position, a 5,000 cycle low-pass filter was approximated. (This was roughly the response of "good" consumer equipment of the time.) An acoustically transparent curtain was placed in front of the barrier, so that listeners could not see what transpired behind it. A six piece dance band playing at a moderate level (about 70 dB average) was on the other side. The vanes were opened or closed at about 30 second intervals. The listeners saw an indicator light change from "A" to "B" in synchronization with the vane changes and were asked which condition they preferred. The tests were repeated, using spoken voice as the sound source, all other conditions the same. The results were conclusive: the great majority of listeners preferred wide-range response under the conditions of the test.
The acoustic/electric test
Olson followed up this acoustical test with a second experiment. The same listening room was used, but this time, the musicians performed in an anechoic chamber where they were picked up in stereo by high quality microphones, with the signal delivered to the listening room by amplifiers and speakers developed by RCA's laboratory. This was the "A" condition. In the "B" condition, roll-off filters in the treble and bass were inserted in the reproduction chain to simulate the response of consumer equipment. Again, the results were conclusive: the great majority of listeners preferred wide-range response.
From today's vantage point, the outcome of Olson's experiments might seem intuitively obvious. But in the late 1940's, the answers to the questions he asked were not so obvious, given the small portion of the population that had ever heard high quality sound reproduction. Olson's experiments were important because RCA was, at that time, an enormously influential company. If RCA's executives had not supported the development of high fidelity recording and playback, the industry as a whole would not have moved as confidently in that direction.
(Olson's account of his tests, and a great deal more, can be found in his book, "Music, Physics and Engineering", second edition, Dover Books, NYC, 1967 (paperback), ISBN-486-21769-8. I recommend this book to all students of the art of recording.)
12. "The Benny Goodman Story" is available on CD, Capitol Jazz CDP 7243 8 33569 2 8, issued 1995. For comparison I recommend Goodman's recordings from 1936 through 1938 for RCA. These are available in several collections as of this writing. My source is a double-CD set on French RCA, "Jazz Tribune No.47, The Indispensable Benny Goodman, Vol.3/4 (1936-1937)", RCA 66470-2 07863. It was available, last I looked, at HMV records in NYC.
13. I have seen contradictory accounts of this early stereo recording system. According to John Pfeiffer's notes for "The Age of Living Stereo" (see note 14) the system had two separate grooves to carry left and right information. However, in the June 1982 issue of Audio, Edward Tatnall Canby refers to the Bell Labs stereo system as "45/45 stereo discs", that is, the same as the later stereo LP in which a single groove carries separate information on each groove wall. According to John Alvin Pierce in his article on the late 1930's work at Harvard (Audio magazine, March 1991), the Bell engineers produced a vertical cut (hill-and-dale) system for their transcription disk because they believed it produced better fidelity. As this is not consistent with a 45/45 cutting system, which produces a lateral cut from the mono portion of the signal, I would be inclined to believe Pfeiffer's account.
14. The best overview of Pfeiffer's work in early stereo is a two CD set, "The Age of Living Stereo: A Tribute to John Pfeiffer", RCA 09026-68524-2, issued 1996. Included are excerpts from the most important of his early experimental recordings from 1953 on, most of which have never been released in stereo before. (They sound magnificent, by the way.) Pfeiffer himself wrote a history of RCA's program, along with notes on the individual recordings included in the set. He died shortly afterwards, so the album was made into a memorial to him and his work. I highly recommend this set to anyone interested in the history of stereo recording.
copyright 1997, Robert Auld