Modern Taste and Ancient Music: Finding an Appropriate Acoustic for William Byrd s Ave verum corpus

Transcription

1 Modern Taste and Ancient Music: Finding an Appropriate Acoustic for William Byrd s Ave verum corpus Will Howie Graduate Program in Sound Recording, McGill University 1. Introduction Capturing Historical Accuracy Sound is ephemeral. A sound exists only for a brief time the time it takes for the sound source to create it, and the time that sound takes to dissipate within the reflections and reverberation of a given acoustic space. A musical note begins, ends, and we never hear that note again this was the reality of music performance for most of human history, until Thomas Edison created the phonograph cylinder in 1877, thus allowing us to create a record of sound a record of a musical performance. For a great deal of 20 th and 21 st century music, one must only look for a commercial recording to observe the performance practice, the compositional intention, and the applied acoustic space for a given piece. But what about music written before the invention of sound recording? How can one possibly know what the original performances sounded like? Of course, for a number of decades, many scholars, musicians, conductors, instrument makers, and others have been researching period performance practice. One of the results of this field of scholarship is that there now exists a great many commercial recordings, especially since the late 1970s, which self-identify as being historically informed or period performances or some other variation on the theme. The artists on these recordings often use original or replica period instruments that are appropriate to the music in question, and employ performance techniques gleaned from historical treatises, written records, music reviews, and other primary sources of period specific information. Ornoy offers an extensive overview of early music performances from different recordings, comparing

2 parameters such as pitch, intonation and temperament, tempo, rhythmic interpretation, and ornamentation [1]. Sung and Fabian, and Holloway conduct similar comparative analysis of recorded works by Bach and Corelli respectively [2] [3]. Noticeably absent from these studies, however, is any mention of the acoustics of the recorded performances. 2. The Technical Perspective The concept of historically accurate performance spaces is one that is rarely, if ever discussed amongst recording engineers and record producers. In his authoritative Handbook for Sound Engineers, John Eargle begins his list of considerations for choosing a recording venue with: Is the space acoustically appropriate? If too live, can it be partially draped to reduce reverberation time? Not much can be done for a space that has too short a reverberation time, although in some instances large expanses of plywood have been used to cover absorptive seating areas to good effect. Some engineers have covered seating areas with plastic sheeting, with some improvement in sound reflectance at high frequencies. [4, pp. 235] Hugonnet and Walder discuss key factors to consider when selecting a recording venue. They define the types of recording venues as: [5, pp ]: 1. Premises specially designed for sound pick-up and recording 2. Acoustically renowned halls 3. Venues with renowned acoustics, but ill-adapted to the type of work performed 4. Multipurpose hall with more or less acoustic treatment 5. Premises with no acoustic treatment 6. Outdoors

3 Notice a complete lack of discussion of the historical accuracy of the recording venue as it relates to the piece(s) being formed. Few commercial recording projects attempt to deliver a historically accurate acoustic to the listener, Tom Beghin s Haydn Project [6] and Sir John Eliot Gardener s Bach Cantata Pilgrimage [29] being two notable exceptions. For acoustic music recording, recording engineers and music producers prize venues known for clarity and sound support: rooms where the early reflections bring focus to musical lines and expand the dimensions of the sound stage, with diffuse late reflections (reverberation) that give a sense of warmth and envelopment to the listener. This all leads to an interesting question: given a specific genre (or piece) of music, have contemporary listeners come to understand or expect a certain type of acoustic as being appropriate an acoustic that may have much more to do with the aesthetic decisions of engineers and producers than with any historical performance practice? 3. Historical Perspective To answer the above question, one could create a simple listening test, comparing known historical acoustic spaces with contemporary recording venues. But what about music where the original performance venue remains a mystery? Such is the case for a great deal of the polyphonic vocal music composed by William Byrd. Byrd (c. 1539/ ) was the most celebrated English composer of his time Grout and Palisca refer to him as the last of the great Catholic church composers of sixteenth century. [7] Byrd, along with Thomas Tallis (who died in 1585), was held in such high regard as to be granted an exclusive license for the printing and publishing of music in England by Elizabeth I. One of Byrd s greatest undertakings, both as a composer and publisher, was his two volumes of Gradualia, originally published in 1605 and 1607 respectively. The contents of the Gradualia, totalling 109 pieces of Latin polyphonic vocal

4 music, include the complete Catholic mass propers for the Sundays and major feasts of the church year (introit, offertory, communion, and some combination of gradual, alleluia, tract, and sequence, as well as some additional works that could have been performed during mass, such as Ave verum corpus) [8]. This massive compositional undertaking is even more astounding given that Byrd wrote (and published) this music at a time when Catholicism was illegal in England. A survey of Michael Greenhalgh s comprehensive Byrd Discographies from and shows that Ave verum corpus, from the 1605 Gradualia, is far and away Byrd s most recorded piece of music [9] [10], making it, by default, his most famous composition for today s listening audience. An examination of nine of these recordings finds that most have been captured in spaces recording engineers would tend to define as a Large Church (reverberation time of 3-6 seconds [5]) or Cathedral (reverberation time greater than 6 seconds [28]) [11, 12, 13, 14, 15, 16, 17, 18, 19]. One of these albums is even titled, Music for Great Cathedrals [19] hilarious, given that it likely would have been impossible for this piece to have been performed in a cathedral during Byrd s time (at least in Protestant England). Byrd had powerful patrons who protected him from persecution due to his religious beliefs Elizabeth I herself intervened and halted his prosecution for Catholic activities in 1592 [20], but this protection would not have extended to most of the English recusant community of the time. Where then would or could have this music been performed? Harley [21], Kerman [22], and McCarthy [8, 23] all offer evidence that clandestine Catholic services were a regular occurrence in Elizabethan England. These services would typically be held in large manor houses, some of which possessed private chapels [24]. All three authors

5 make reference to a famous account of a rather impressive private chapel built by one Lady Montague: Here almost every week was a sermon made, and on solemn feasts the sacrifice of the mass was celebrated with singing and musical instruments, and sometimes also with deacon and subdeacon. And such was the concourse and resort of Catholics, that sometimes there were 120 together, and 60 communicants at a time had the benefit of the Blessed Sacrament. [22, p. 49] Such a large and open affront to the authorities could not have been the norm of the time. Perhaps a more representative account is that of Henry Garnet, who in 1605 describes a clandestine English Catholic celebration: We kept Corpus Christi Day with great solemnity and music, and the day of the Octave made a solemn procession about a great garden, the house being watched, which we knew not until the next day [22, p. 50] We are now starting to get a picture of the types of spaces where Ave verum corpus might have been performed: domestic, probably not very large (with the exception of Lady Montague s chapel), and, one would assume, not particularly resonant or reverberant a far cry from the acoustic presentation typical on most commercial recordings. In Sacred Songs in the Chamber, Milsom takes things a step further, showing that while Byrd s earlier Latin motets could have found a place in Anglican church services (using either the original text, or an English translation), the motets of the Gradualia are totally absent from the books used by church choirs. [25, p. 177] Milsom finds a pattern of private ownership for Byrd s later publications of Latin-texted music, arguing a very low likelihood that any of this music would have been performed by Anglican church choirs

6 [25]. Milsom concludes with a plea to rethink our current, favoured performance (and by extension recording) practices for Byrd s Gradualia: If on the other hand it is the music we put first, then we can look forward to the day when works with a recent history of choral interpretation finally break free from performance styles for which they were never intended, and are returned to the chambers where they more legitimately belong. [25, p. 179] The modern performance aesthetic of Byrd s Latin motets has been defined, by and large, by the 19 th and 20 th century choirs who, able to perform this music without threat of arrest, brought pieces like Ave verum corpus to their home bases of performance: large churches and cathedrals the very same spaces prized by recording engineers and record producers for their spacious, beautiful acoustics. In order to take up Milsom s challenge, and bring Ave verum corpus back to the chambers where they more legitimately belong [25], we would first need a solid understanding of what these spaces were like, and how they sounded. Unfortunately, as McCarthy puts it: We can only guess where, when, and why Byrd s piece [Ave verum corpus] was sung in the English recusant community. [8, p. 19] McCarthy later adds, The problem only gets worse in a specific study of recusant liturgical practice. The exact details that would be most useful the times and places where Mass was celebrated, the people who took part, their connections and circumstances would have been the most incriminating of all, and therefore tend to be the most obscure. [8, pp ] So though we may have some idea of where Ave verum corpus was performed during Byrd s time, little to no evidence has been offered to describe what the acoustics of such spaces would have been like. Details such as building

7 construction, furnishings, number of people present, clothing (and its absorptive coefficient) are sadly lacking from the current literature. 4. Experimental Design and Implementation In order to further examine the question of acoustical appropriateness according to a 21 st century audience, a listening test was designed and implemented. Given its fascinating lack of known original performance space, as well as its popular place in the current choral music repertoire, Byrd s Ave verum corpus was selected as the musical stimulus. The piece was performed and recorded in the Spatial Audio Lab at the Centre for Interdisciplinary Research in Music Media and Technology, at McGill University [Fig 1]. The Spatial Audio Lab is a semianechoic space, with a measured reverberation time (T30) of 110ms, thereby falling within the definition of acoustically dead. The piece was performed by four professional singers (see Acknowledgements) all of whom have considerable experience singing 16 th -century Latin motets. The sound was captured using two Schoeps MK2S omnidirectional microphones, set at a height of 1.73m, spaced 59cm apart, at a distance of 1.7 meters from the singers (capsules oriented towards the singers heads). Both microphones were routed to the internal preamps and A/D converters of an RME Fireface UFX, and recorded to Pro Tools 12.1 at a resolution of 96KHz/24bits.

8 Figure 1. Spatial Audio Lab Having captured Ave verum corpus in a space with no acoustic imprint of its own, a number of different acoustics could then be applied to the performance. This was accomplished in Pro Tools, using Audio Ease s Altiverb, a commercial convolution reverb plugin featuring an extensive library of impulse responses of various acoustic spaces. (These spaces are measured with no audience present.) Based on the previously discussed survey of recordings of Ave verum corpus, as well as what is known about the types of performance spaces potentially associated with the piece, five contrasting venues (ranging from no reflections to highly reverberant) were selected to be applied to the semi-anechoic recording. 1. None /Anechoic (Reverb Time: 110ms [Fig. 1] 2. Castle Zuylen, Meeting Room - Netherlands (Reverb Time: 1.3) [Fig. 2] 3. St. Mary s church Coddenham, UK (Reverb Time: 2.1) [Fig. 3] 4. Utrecht Music College, Chapel Netherlands (Reverb Time: 4.0s) [Fig. 4] 5. King s College Chapel Cambridge, UK (Reverb Time: 12s) [Fig. 5]

9 Figure 2: Castle Zuylen, Meeting Room. Note the wood floors, plaster walls and wood ceiling. Figure 3: St. Mary s church. A small stone church with a short reverb time.

10 Figure 4: Utrecht Music College, Chapel. Figure 5: King s College Chapel. Completed in mid 1530s. Two contrasting excerpts of Ave verum corpus were then selected to be used as musical stimuli. The first excerpt, measures 1-4, features a largely declamatory, homorhythmic texture. The second excerpt, measures 36-43, has a densely polyphonic texture, including a number of imitative entries and modules. Each excerpt was under 30s long. While mixing the stimuli, great care was given to ensure a similar balance of direct sound to reverberant sound for each acoustic condition the ideal always being the perspective of a listener at the critical distance (equal balance of direct and reverberant sound) [5]. All ten stimuli (5 acoustic conditions x two musical

11 excerpts) were level matched using a combination of a LUFS meter on the mix bus, and an SPL meter at the listening position. Final adjustments for the levels were then made by ear, by three professional recording engineers. A listening test was then created using Pro Tools. The test took place in McGill University s Studio 22, an acoustically treated control room that fulfills ITU-R BS.1116 requirements [26]. Dynaudio BM15 loudspeakers were used for audio playback. For a given trial, an excerpt would playback on a repeating loop, with subjects able to freely switch between the different acoustic conditions, labelled A, B, C, D, and E [Fig. 6]. Labelling and ordering of the different acoustic conditions was randomized per trial. For each musical excerpt, subjects were asked to select the acoustic condition they felt was the most appropriate for the music being performed, and to provide a brief reason why they had made their selection. After completing both trials, subjects filled out a short demographic survey. A total of 23 subjects performed the listening test. Of those, 16 self-identified as trained recording engineers. 7 indicated they were taking part in a graduate seminar on 16 th -century motets. 13 identified as having significant choral singing experience and 6 as having significant experience singing 16 th /17 th -century Latin motets [Appendix A]. Figure 7: Testing interface in Pro Tools.

12 5. Results and Discussion For both musical excerpts, Utrecht Chapel was selected as the most appropriate acoustic condition by a clear majority of subjects, 74% for excerpt 1(homorhythmic) and 70% for excerpt 2 (imitative texture) [Fig 8 and 9]. A strong majority of subjects (83%) chose the same acoustic condition regardless of the musical excerpt. The above results appear to be due primarily to the large influence of the trained recording engineers cohort on the data, 88% of whom selected Utrecht Chapel as the most appropriate acoustic [Fig. 10]. Additionally, all subjects in the recording engineer cohort chose the same acoustic condition for both musical excerpts. None of the subjects chose the anechoic condition as an appropriate acoustic, regardless of the musical excerpt. This is not surprising: Toole [27] and many others have pointed out that humans are not accustomed to anechoic spaces, and are shown to be more comfortable listening to speech and music in enclosed spaces with audible reflections. Most, if not all listeners, trained or otherwise, would find the sound of the Anechoic condition quite jarring. EXCERPT 1 ACOUSTIC CHOICES King's 4% St. Mary's 22% Utrecht 74% Figure 8: Subject selection results for Excerpt 1.

13 EXCERPT 2 ACOUSTIC CHOICES Meeting Room 9% King's 4% St. Mary's 17% Utrecht 70% Figure 9: Subject selection results for Excerpt 1. Recording Engineer Acoustic Choices Utrecht Chapel St. Mary's Kings College Chapel Figure 10: Selections for most appropriate acoustic condition made by recording engineers. There was considerably less consistency in acoustic condition selection among the non-recording engineer cohort, as can be seen in the below table: Subject Example 1 Acoustic Choice Example 2 Acoustic Choice Motet Seminar? 16 th c. Motet Singing Experience? 5 Utrecht Chapel Utrecht Chapel Yes No 6 St. Mary s St. Mary s No No 8 St. Mary s Utrecht Chapel Yes No 17 Utrecht Chapel Meeting Room Yes Yes 19 Utrecht Chapel St. Mary s No Yes 21 St. Mary s Meeting Room Yes Yes 23 St. Mary s St. Mary s Yes No

14 Note that for this cohort, St. Mary s was the most chosen acoustic condition, overall, though just barely. Also note that over half of the subjects in this cohort chose different acoustic conditions between the two excerpts. A particularly interesting cohort are those with significant experience singing 16 th -century Latin motets. Among those also trained as recording engineers, all three chose Utrecht Chapel for both musical selections, which is in keeping with that cohort s clear preference for the acoustic condition that, frankly, most resembles a typical commercial recording. However, for the nonrecording engineers, all three chose a dryer acoustic for the 2 nd, more imitative textured musical excerpt. Their selection comments are quite revealing: Subject 17: a bit of a ring to it without obscuring things Subject 19: A little less resonant so the echo doesn't interfere with the harmonies Subject 21: Clean enough to hear the different timbres clearly in polyphonic entries Clearly these subject are prioritizing clarity of the polyphonic lines and text comprehension over the potentially warmer and more enveloping (and blending ) acoustics of the more reverberant conditions. It is quite interesting that these sentiments did not manifest in amongst the recording engineers who had the same type of motet performance experience. When examining all the subject decision comments, a clear trend emerges: for those who chose the more reverberant conditions (Utrecht and King s), the most salient qualities were appropriate amount of reverb and clarity/balance of voices. When choosing the less reverberant conditions, subjects tended to focus on clarity of harmonies and polyphony.

15 Surprisingly, only one subject selected King s College Chapel as the most appropriate acoustic condition. After the test, a number of subjects remarked that the reverb tail for King s seemed exaggerated, or even ridiculous (one subject referred to the sound as tasteless ). And yet King s falls perfectly within Meyer s description of typical reverberation times for cathedrals [28]. Judging by Greenhalgh s discographies, Byrd is certainly a popular choice for the choir of King s College. Perhaps North American listeners are simply not accustomed to attending concerts in such highly reverberant spaces or perhaps the acoustic was considered overwhelming for a one-on-a-part performance. 6. Conclusion When discussing a methodology for recording choral music, Eargle writes: Reverberation is important in the texture of a good choral recording. A recording made in a large church will benefit from the natural acoustics of the space [4, p. 254]. It would appear a clear majority of contemporary listeners agree with him. Of all the acoustic conditions, Utrecht Chapel is by far the most representative of the recording engineer s ideal venue: a fairly neutral timbre, a good natural balance of direct to diffuse energy, strong early reflections that give clarity and support to the vocal lines, and an enveloping, tasteful reverb tail that falls within Hugonnet and Walder s suggested reverberation times for a Church [5]. One could argue that this test s majority of trained recording engineers as subjects presents an unnatural bias in the results; however, this is likely not the case. As those who are responsible for the creation of recorded content, it is the recording engineers and record producers who currently shape the aesthetic at large for how recordings of different genres of music should sound. It should come as no surprise then that the

16 acoustic condition closest to the engineer s ideal would be chosen as most appropriate by the majority of subjects (including several non-recording engineers). The Meeting Room condition may have been closest to a historically correct acoustic, but is was one of the least popular choices. (This is exactly why subjects were not instructed to specifically think about potential historical accuracy when making their decisions the test did not want to introduce such a bias.) Though Ave verum corpus may have been performed predominantly in small, relatively dry spaces during Byrd s lifetime, it can be said for certain that most contemporary listeners do not want to hear the piece in those acoustic conditions. One last thing to consider: as can be seen in Figure 5, Utrecht Chapel is not an overly large space, with a floor plan of only 14m by 6m, seating an audience of around people. Although difficult to know from the current literature on Byrd, it is not unrealistic to assume that some of the wealthier English Catholics during his time may have had private chapels similar in dimensions and construction to Utrecht Chapel. As such, in the results of this listening test, an interesting intersection point has been found between a 21 st century acoustic preference, seemingly based on modern sound recording practices, and a viable historically accurate acoustic condition for one or more of the original performances of William Byrd s Ave verum corpus. 7. Acknowledgements This research was supported by the Social Studies and Humanities Research Council, and the Centre for Interdisciplinary Research in Music Media and Technology. Special thanks to the four singers who participated in the anechoic recording: Rona Nadler (soprano), Meagen Zantingh (alto), Matthew Milner (tenor), and Dave Benson (bass).

17 8. References [1] E. Ornoy, Between Theory and Practice: Comparative Study of Early Music Performances, Early Music, vol. 34, no. 2, pp , May [2] A. Sung and D. Fabian, Variety in Performance: A Comparative Analysis of Recorded Performances of Bach s Sixth Suite for Solo Cello from 1961 to 1998, Empirical Musicological Review, vol. 6, no. 1, pp , Jan Web access: Accessed: :00 UTC [3] J. Holloway, Corelli s Op.5: Text, Act and Reaction, Early Music, vol. 24, no. 4, pp , Nov Web access: Accessed: :41 UTC [4] J. Eargle, Handbook for Recording Engineers. New York: Van Nostrand Reihold Company, [5] C. Hugonnet and P. Walder, Stereophonic Sound Recording: Theory and Practice. New York: Wiley, [6] W. Woszczyk et al, Recording multichannel sound within virtual acoustics, in AES 127 th Convention, New York, USA, 2009, pp [7] D. J. Grout and C. V. Palisca, A History of Western Music. New York: W. W. Norton & Company, Inc. 1996, pp [8] K. McCarthy, Liturgy and Contemplation in Byrd's Gradualia. New York: Routledge, [9] R. Turbet, William Byrd: A Guide to Research (Second Edition). New York: Routledge [10] R. Turbet, William Byrd: A Guide to Research (Third Edition). New York: Routledge [11] S. Darlington et al., Oxford Church Anthems. Charlottesville, VA.: Nimbus, [12] W. Byrd et al., Ave verum corpus: motets and anthems. Cambridge, England: Collegium Records, [13] W. Byrd et al., Mass for five voices; Mass for four voices; Mass for three voices; Ave verum corpus. Oxford: Gimell, [14] W. Byrd et al., English Renaissance. New York: RCA Victor, [15] J. Rutter et al., Faire is the heaven: music of the English Church. Cambridge, England: Collegium Records, 1988.

18 [16] T. Brown, Clare College Choir, Tudor Anthems and Motets. Heritage, [17] G. F. Handel et al, King s College Choir: England My England. Hong Kong: Naxos Digital Servies US Inc., [18] C. Monks et al., Naked Byrd. Perivale, Middlesex: Signum Classics, [19] M. Stoddart et al., Music for Great Cathedrals. Hong Kong: Naxos Digital Services/The Gift of Music, [20] Kerman, Joseph, and Kerry McCarthy. "Byrd, William." Grove Music Online. Oxford Music Online. Oxford University Press 7. Accessed: :00 UTC [21] J. Harley, William Byrd: Gentleman of the Chapel Royal. Brookfield: Scholar Press [22] J. Kerman, The Masses and motets of William Byrd. Berkeley: University of California Press, [23] K. McCarthy, Byrd. New York: Oxford University Press, [24] F. Hudson, The performance of William Byrd s church music. I. Music acceptable in the Anglican Church, American Choral Review. vol. 12, pp , [25] J. Milsom, Sacred songs in the chamber, English choral practice, J. Morehen, Ed. Cambridge: Cambridge University Press, 1995, pp [26] "Methods for the Subjective Assessment of Small Impairments in Audio Systems Including Multichannel Sound Systems," ITU-R Recommendation BS , International Telecom Union: Geneva, Switzerland, 1997, pp [27] F. E. Toole, Loudspeakers and Rooms for Sound Reproduction A Scientific Review, Journal of the Audio Engineering Society, vol. 54, no. 6, pp , June [28] J. Meyer, Acoustics and the Performance of Music. New York: Springer, [29] B. Kok and E. Groot, "A Mobile for Classical Recording," in AES 112th Convention, Munich, Germany, 2002, pp

19 Appendix A: Additional Charts Subject Demographics Total Participants 23 Significant 16th/17th c. Motet Singing Experience 6 Significant Choral Singing Experience 13 MUHL 681 Participant 7 Trained Recording Engineer