Lost Voices of Hagia Sophia

For many critical listeners, and not just readers of this magazine, the absolute sound—the sound of live, unamplified music in a real space—is a touchstone of the audiophile pursuit. Jazz recorded at a repurposed pawnshop in Stockholm, chamber music from a 19th-century bank auditorium in upstate New York, the Cowboy Junkies at a Toronto church, or any number of “Golden Age” orchestral tapings from London, Boston, and Chicago—what these productions have in common is a strong sense of place. The venue chosen for performance or recording can support an understanding of music’s meaning in a powerful way. There is, however, one extraordinary acoustic space that for cultural and geopolitical reasons has been unheard for centuries.

The Church of Hagia Sophia, which translates from Greek as “Holy Wisdom,” was constructed in the Sixth century by the Byzantine emperor Justinian I in Constantinople, now Istanbul. For roughly 900 years, it was the epicenter of the Christian Orthodox faith. For a millennium, it had the largest domed interior of any building on earth and its influence on subsequent architecture was enormous. The interior volume of the nave is 255,800 cubic meters; the cathedral’s dome has a diameter of nearly 32 meters and a height of 56.6 meters—far taller than the largest medieval cathedrals of Western Europe. The reflective surfaces of marble and gold mosaic, the columns of the aisles and galleries, and the sheer size of the room contribute to a reverberation time that is frequency-dependent but as long as 12 seconds. An enveloping sound was a defining characteristic of the religious rites that occurred at Hagia Sophia, which could involve hundreds of clergy plus a congregation of up to 16,000 people, most of whom would have been singing at some point in a service. The aural experience must have been incredible, surely focusing the consciousness of all present on the spiritual realm.

That all changed in May of 1453, when Sultan Mehmed II conquered Constantinople and converted Hagia Sophia to an Ottoman mosque, adding Islamic features inside and four minarets to the exterior. Religion of any kind was banished from Hagia Sophia by Kemal Atatürk in the 1930s and the building is now a World Heritage Site and museum, complete with a gift shop and café. With the increasing Islamization of Turkish society in recent years, readings from the Qur’an have been permitted, though these are amplified and don’t at all exploit the singular acoustics of the space. The prospects of hearing Christian liturgy—medieval Byzantine chant—performed or recorded in Hagia Sophia anytime soon are slim to none.

Dr. Bissera V. Pentcheva is an art history professor at Stanford University with a consuming interest in “aural architecture”—she’s written two books on the subject and edited two others. “My work has always pursued the animation of medieval art, seeking to reconstruct the original conditions in which the art and architecture were perceived,” Pentcheva told me in the course of an email correspondence. “Back in 2008, I was completing a book, The Sensual Icon: Space, Ritual and the Senses in Byzantium, and one chapter explored Hagia Sophia and the role of incense and voice. In doing the research, I stumbled upon acoustic measurements of Hagia Sophia done by the Technische Universität of Copenhagen in 2001–2003. The article stated the long reverberation time, but did not provide any further explanation linking the hard-core data to the cultural framing of reverberant acoustics in Byzantine ritual and spirituality.” 

Around that time, Dr. Pentcheva met Jonathan Abel, a consulting professor at Stanford’s Center for Computer Research in Music and Acoustics (CCRMA, pronounced “Karma”). Dr. Abel’s research interests include the analysis and reproduction of acoustic environments—his more than 100 publications include “VampireVerb: A surreal simulation of the acoustic of Dracula’s castle.” Pentcheva passed along the Danish paper and Abel’s interest was piqued. “Three main features characterize reverberation—the reverberation time, echo density profile, and ‘wet’ equalization,” he explained. “The published data gave us only the reverberation time and a suggestion of the ‘wet’ level, but some work we had been doing at CCRMA gave us confidence that we could extract a rather complete picture from a balloon pop recording.” Drs. Pentcheva and Abel became co-directors of Icons of Sound, a multi-disciplinary project devoted to understanding the acoustic of Hagia Sophia from both aesthetic and engineering standpoints.

In May and December of 2010, Dr. Pentcheva was permitted access to Hagia Sophia during hours it was closed to the public to make a total of four balloon pop recordings. Equipped with two small microphones worn near her ears and a portable digital recorder, Pentcheva positioned herself on the western side of the nave as an assistant popped balloons under the dome with a pin attached to a long, thin pole. The resulting recordings were brought back to California, where Abel used them to develop a set of “impulse responses” and a convolution-based “auralization” algorithm to simulate Hagia Sophia’s acoustic. The methodology was validated by repeating the balloon pops in a smaller but similarly shaped local structure, the Stanford Memorial Church, along with sine wave sweeps and noise bursts played through high-quality loudspeakers—these sounds recorded with stand-mounted omnidirectional and cardioid microphones. There was good agreement between the simulation and the actual recording; the tools were in place to perform Byzantine chant in an artificially rendered version of the Turkish venue’s acoustic.

Cappella Romana is a professional vocal ensemble based in Portland, Oregon, founded in 1991 by Alexander L. Lingas. The group performs sacred music from the medieval period as well as contemporary repertoire with a similar devotional character, works by composers like Arvo Pärt and John Tavener. Lingas, now a Professor of Music at City, University of London, is a Byzantine chant authority and has conducted a great deal of scholarly research in the field, including investigations into liturgical practice at Hagia Sophia. 

 Dr. Lingas became a key member of the Icons of Sound team and prepared programs of Byzantine music for performance at Stanford’s Bing Hall in 2013 and 2016. Cappella Romana members wore microphones and their vocalizations were processed with Abel’s auralization algorithm. This allowed the 15 singers—and an audience—to experience their performance in real time as it would have been heard in the vast space of Hagia Sophia by playing their singing back through 30 loudspeakers deployed throughout the hall. Electronic techniques assured that Bing’s acoustic wasn’t superimposed on that of Hagia Sophia. The concerts were an unqualified success. 

Spectrographs comparing the “dry” sound captured by the head-mounted mics to the “wet” sound resulting when that signal is processed with the Hagia Sophia convolution algorithm demonstrate objectively the substantial effect of the venue on singing. Figure 1 shows the same brief passage—the first approximately six seconds feature a male soloist and the second six the full choir—before and after auralization. Quiet signals are in black to magenta and loud ones in yellow. The “dry” tracing shows silence between words and blurring when the church acoustic is applied. Just as importantly, Hagia Sophia changes the frequency component of the sound at a given moment and introduces an element of amplitude variation that a listener can perceive as “movement.”

To make a recording documenting their accomplishment, Lingas, Cappella Romana, and Abel returned not to Bing but to CCRMA to employ the auralization technique in a studio environment. The singers had small Countryman B6 microphones taped to their foreheads; these fed four Adam AX-77 powered loudspeakers. “Dry” output from the singer’s microphones plus an array of eight omni and cardioid mics deployed in the room were used to make the recording. HD and multichannel programs were mastered at Skywalker Sound, the CD at Perfect Record in St. Paul. 

The mix is “synthetic” in that the goal is to present the experience of hearing the music from a specific location in the room. Alexander Lingas explained: “The perspective of the recording is that of the part of the congregation that is closest to the action, being slightly further into the nave (away from the apse) but still under the dome,” he told me. “The choir is around the ambo [a raised platform for readings and homilies], with soloists ascending it for certain pieces. When a Deacon is heard, he is at the doors of the sanctuary. Celebrants from the higher clergy are inside the sanctuary, either in front of the altar (singing into the apse) or, in the case of one blessing, pronounced from the throne in the center of the apse itself…. In a couple of tracks there are congregational responses and you can hear the sound coming from behind as well as in front, giving the listener the sense of being surrounded by singing congregants.” Dr. Lingas continued: “These placements were determined in postproduction sessions for which I provided plans based on my academic work on the Constantinopolitan cathedral rite. I think that the result, which tallies with my experience chanting modern Orthodox services in many buildings, some with very resonant acoustics, gives the listener a sense of the hierarchical soundscape of Byzantine worship, at least as experienced by people privileged to occupy prime spots. The experience of mothers with young children up in the galleries of Hagia Sophia would have been something different again.”

The performance of—and even the consumption of—the earliest surviving “classical” music is very much the domain of specialists. Listeners devoted to Bach, Schubert, Mahler, and Shostakovich usually have little currency with medieval repertoire, their last exposure possibly the music history course they took as undergraduates. Some understanding of religious practice and historical context is necessary and Alexander Lingas contributes helpful notes for this release on Cappella Romana’s own label. The Byzantine rite of the time, Dr. Lingas explains, comprised three components—the Communion Service (or “Divine Liturgy”), the “stationary liturgy,” a system of processions, and the “Divine Offices,” a cycle of services occurring at specific times of day such as Matins (at dawn) or Vespers (at sundown). Lingas includes all three elements in constructing a program that references what was one of the most important events of the liturgical year at Hagia Sophia, The Exultation of the Holy Cross. Employing Greek manuscripts as his sources, the 77-minute program includes solo and responsorial chant, psalmody, and hymns.

Compared to most modern performances of its Western European counterpart, Gregorian plainchant, Byzantine chant sounds less chaste and austere. The music has a sensual, ecstatic quality with a Middle Eastern inflection that features searching, sinuous melody. That melody is monophonic—just a single line—but the presence of an inexorable drone intoned by several low male voices produces the potential for harmonic tension. There are lengthy melismatic passages—a single word of text is sung on multiple notes—and the vocalization of sounds with no meaning at all (teretismata). The music is very eventful with an energizing emotional trajectory. The closing Asmatikon Cherubic Hymn is typical, a 13-minute piece that evolves from a place of mystery and awe with a tonal center of E minor and the frequent occurrence of minor second intervals that, after increasingly extravagant excursions of the melodic line, arrives at a triumphant conclusion in something like C major.

Lost Voices of Hagia Sophia is available as a 2-disc set, a CD plus a Blu-ray disc that holds high-resolution stereo and 5.1 programs plus ten-channel Auro-3D and Dolby Atmos versions, each including four speakers at elevation. Also included on the BD is an expertly produced 24-minute documentary film that details the Icons of Sound project. The sound is extraordinary. One might expect an undifferentiated mush dominated by low frequencies, but this isn’t the case at all. Pentcheva, in her liner note, describes “a ‘waterfall’ of golden, glittering sound” that she attributes to the church’s dome being “especially reflective of high-frequency sound.” Particularly in 24-bit multichannel, the long reverberant tail is completely believable as deriving from the initial sound produced by the singers; in stereo, the reverberation is still convincing, though there’s some loss of a sense of the vastness of the space. The Red Book version on the CD is still an involving listening experience even if the  “largeness” of the acoustic is more generic and the magically clarifying sparkle of sound as it descends from the cupola is missing to a degree. I’m not set up for 5.1.4 playback; both Alexander Lingas and Jonathan Abel noted that there’s plenty of relevant height information to be communicated. Some of that comes through with 5.1 surround.

Pentcheva and Abel have started applying these techniques to a church in Conques, France, a popular stop for pilgrims travelling the “Way of St. James” to Spain. The balloon pop recordings are done and the impulse-response calculations completed. Live auralization recordings were to begin in May 2020, but the COVID-19 pandemic has necessitated a delay. What Icons of Sound achieved with the Hagia Sophia project leads me to expect that this singular joining of scholarship, musicianship, and advanced audio technology will continue to illuminate the significance of music from the distant past. Bissera Pentcheva commented that her study of aural architecture has made her “recognize how the act of singing, and especially singing in a reverberant space produces an invisible aural energy that envelopes the congregation and makes the divine present as an acousmêtre, a bodiless voice. This process draws attention to breath/Spirit and its role in inspiriting bodies, thus transforming them into images of God.”