Icons of Sound: Auralizing the Lost Voice of Hagia Sophia
2017; University of Chicago Press; Volume: 92; Issue: S1 Linguagem: Inglês
10.1086/693439
ISSN2040-8072
AutoresBissera V. Pentcheva, Jonathan S. Abel,
Tópico(s)Ancient Mediterranean Archaeology and History
ResumoPrevious article FreeIcons of Sound: Auralizing the Lost Voice of Hagia SophiaBissera V. Pentcheva and Jonathan S. AbelBissera V. Pentcheva and Jonathan S. AbelBissera V. Pentcheva, Stanford University ([email protected])Jonathan S. Abel, Center for Computer Research in Music and Acoustics (CCRMA) at Stanford University ([email protected])PDFPDF PLUSFull TextSupplemental Material Add to favoritesDownload CitationTrack CitationsPermissionsReprints Share onFacebookTwitterLinked InRedditEmailQR Code SectionsMoreThe interdisciplinary project Icons of Sound (2008–present) is codirected by Bissera Pentcheva (Department of Art and Art History) and Jonathan Abel (Center for Computer Research in Music and Acoustics [CCRMA]) at Stanford University.1 Bridging humanities and exact sciences, this research focuses on Emperor Justinian's sixth-century church of Hagia Sophia, uncovering the synergy that once existed among acoustics, chant, and aesthetics (Fig. 1). The liturgical rite, celebrated within this monument from late antiquity until at least the Fourth Crusade in 1204, was known as the ekklēsiastēs ("cathedral") or more commonly referred to as the asmatic, or sung, rite.2 As a museum today, the interior of Hagia Sophia is off limits for any performance involving the human voice; this ban pertains both to religious and artistic initiatives. Digital technology has become our only means to restore and experience the lost voice of the Great Church. Icons of Sound has collected acoustic data on site using inflated balloons popped in the interior. Extracting from these samples the impulse response of the space, the team has successfully imprinted Hagia Sophia's acoustic signature on live performance of Byzantine chant, first in a studio setting in 2011 and then at Stanford's Bing Concert Hall in 2013 and 2016.3 This article presents the method and the results of this research.Fig. 1. Hagia Sophia, 532–37 and 562, interior (The Byzantine Institute and Dumbarton Oaks Fieldwork Records and Papers, c. late 1920s–2000s, MS.BZ.004, Image Collections and Fieldwork Archives, Dumbarton Oaks, Trustees for Harvard University, Washington, D.C. Photograph: Courtesy of Image Collections and Fieldwork Archives, Dumbarton Oaks, Trustees for Harvard University, Washington, D.C.)View Large ImageDownload PowerPointThe Acoustics of Hagia SophiaThe current building of Hagia Sophia dates to the time of Emperor Justinian (527–65). The cathedral, built in a swift campaign in 532–37, had an original flat dome supported on pendentives that created a luminous interior. Damaged in an earthquake in 557, this cupola collapsed and was then rebuilt in the period 558–62 with a higher apex. It rises 56.60 meters above the floor and has a maximum diameter of 31.87 meters (Fig. 1).4 The height of this space thus exceeds by far the tallest of medieval cathedrals of Western Europe.5 The vast interior volume of the nave is 255,800 m3 and it can house roughly sixteen thousand people.6 The Justinianic Hagia Sophia thus presents a grand stage for public ceremony and religious ritual. Hagia Sophia's long reverberation time of more than 10 seconds results from the immense interior volume and the reflective surfaces of marble and gold mosaic.7 The reverberation time remains uniform across a vast spectrum of frequencies in the range of the singing human voice, from 200 Hz–2,000 Hz (Fig. 2).8 It reaches a peak of 12 seconds at frequencies in the range 250–500 Hz, and it reduces to 10 and 8 seconds respectively for frequencies of 1 and 2 kHz.9 The colonnades of the aisles and galleries that frame the naos create coupled spaces. This characteristic, together with the architectural detail and marble revetments, enables the reflected sound energy to mix quickly and fill the interior with diffuse reflections. This enveloping sound constitutes a characteristic of some of the best concert halls today. Yet the duration of the reverberation in the Great Church is extremely long, and this feature sets it apart from modern performance spaces, whose reverberation time is by contrast often less than 2 seconds.10 Because of this, clarity and intelligibility of speech vocalization in Hagia Sophia is poor overall.11Fig. 2. Reverberation time (T30) averaged over eight positions measured with broadband sources in Hagia Sophia (© Wieslaw Woszczyk)View Large ImageDownload PowerPointThe long sustain of the late field reverberation and enveloping sound result in room acoustics well suited for monodic chant.12 The interior particularly prolongs high-frequency harmonics. The choristers can use the space as a musical instrument that gives body and fullness to their voices. They can interact further with the harmonics sustained by the space and create a form of polyphony emerging from the synergy of the singer's output and the bright and sustained return of the building.The interior creates several significant acoustic effects. The first is amplification, as the sung notes are held within the interior for a long time. A gradual buildup emerges as more sound energy is continually added and only slowly absorbed. This phenomenon produces a fuller and richer sound. The second effect is that of overlapping and dissolving of notes. The building responds to the monody by sustaining high-frequency returns, which interact and may gradually evolve from dissonant to consonant harmonics. The third phenomenon is that of acoustic waterfall. Raised almost fifty-seven meters above the floor, the cupola reflects and scatters the sound waves over a much wider area of the floor.13 The dome is especially reflective of high-frequency sound, reinforcing these particular harmonics in the interior. The shape of the cupola can both concentrate and scatter the sound energy. This phenomenon stirs the synesthetic effect of aural and optical brightness as it combines the acoustic reflection with the visual reflection of light off the gold mosaic surface of the dome. As a result, the sonic and visual brightness acts as a mirror reflecting the imagined splendor of the angelic choirs.14 The fourth phenomenon is lack of clarity. Melodic progressions are smeared over time; they collide tonally or harmonize with the newly sung pitches, increasing the perceived loudness of the sound but decreasing the clarity of the vocalization.Measuring the Acoustics of Hagia SophiaThe scientific measurement of Hagia Sophia started in 2000–2003 with the CAHRISMA (Conservation of Acoustical Heritage by the Revival and Identification of Sinan's Mosque Acoustics) Project, led by a research team from the Technical University of Denmark; the group included the acoustic engineers Jens Rindel, Claus Christensen, Anders Gade, and Christoffer Weitze.15 Icons of Sound commenced measuring in Hagia Sophia in 2010. Concurrently with our research, Wieslaw Woszczyk from McGill University conducted a third independent set of measurements in 2013.16Stanford's Icons of Sound's first acoustic measurements were based on balloon pops.17 Our goal was to obtain from this data the impulse response of the space. "Impulse response" refers to the acoustic signature of a space or the sound produced in response to a transient input sound, such as a pistol shot or a balloon pop. An impulse response can be likened to an indexical "snapshot," which records the acoustic signature of a space measured as the decay of –60dB in seconds of an impulsive signal fired in that space. A broadband sound source spanning the frequency range of human hearing (20 Hz to 20 kHz) is typically used to give a comprehensive "picture" of the way the particular space imprints itself on sound.What we needed to analyze and to simulate the acoustics of Hagia Sophia was a set of impulse-response measurements. There are many methods available to measure room impulse responses, most commonly involving playing test signals, such as swept sinusoids or Golay codes, into the space from a loudspeaker and recording the room's response at a set of microphones. The drawback of this approach is that it requires a significant amount of time to set up and tear down the equipment in order to make the measurements. Since access to Hagia Sophia is very limited, these approaches were not available to us. In our work, we circumvented the logistical difficulties associated with loudspeaker-based measurement by relying on balloon-pop responses, recorded by Bissera Pentcheva in May and December 2010 in Hagia Sophia (Fig. 3).18 This approach calls for only a handheld recorder, a set of omnidirectional microphones that can easily be worn on the body (clipped in the hair near or around the ears), and a balloon.Fig. 3. Hagia Sophia, Istanbul, balloon-pop response (above) and associated spectrogram (below), December 2010 (diagram by Jonathan S. Abel)View Large ImageDownload PowerPointBalloon-pop responses are not impulse responses, however, and we developed a method to convert our balloon-pop recordings into impulse responses (Fig. 4).19 This method was unproven, and to test its effectiveness we conducted an experiment in Stanford's Memorial Church. We worked with a student, Konstantine Buhler, trained in Orthodox chant. He stood and sang in position B; our room microphones were set in position C (Fig. 5). We recorded Konstantine first "dry" (his microphone was set close to his mouth so that it collected the sound with minimal room acoustics, that is, anechoically). Then we recorded him "wet" (the microphones set in position C in the nave and away from the singer recorded sound that contained the acoustic imprint of the room). Leaving a microphone at position C, we popped several balloons from position B, where Buhler had been singing, and recorded the responses. We converted the balloon-pop recordings into room impulse responses using our method and then convolved the estimated impulse responses with the dry recording to obtain a simulated Memorial Church response at position C to Buhler's singing at position B.Fig. 4. Hagia Sophia impulse response (above) and associated spectrogram (below), recovered from the December 2010 balloon-pop response recording (diagram by Jonathan S. Abel)View Large ImageDownload PowerPointFig. 5. Memorial Church, Stanford University, floor plan (plan by Mcginnly, licensed under CC BY-SA 3.0 via Commons, https://commons.wikimedia.org/wiki/File:Stanford_Memorial_Church_Plan.jpg#/media/File:Stanford_Memorial_Church_Plan.jpg)View Large ImageDownload PowerPointThe spectrogram of the simulated position C signal is plotted (Fig. 6, below) on the same time axis as the spectrogram of the signal actually recorded at position C (Fig. 6, center). The spectrograms and sound of the recorded and simulated signals are very similar in the mid-frequencies and above, verifying the effectiveness of the room simulation via impulse response derived from balloon-pop measurements. Where they differ is in the low frequencies, where the church heating system (which could not be turned off) added noise to the recording. It was an unexpected benefit of the simulation that, compared to the actual recording, stray environmental noises could be eliminated.Fig. 6. Spectrograms of chant recorded in Memorial Church, Stanford University, headset microphone (above), nave microphone (center), and simulated nave microphone (below) (diagram by Jonathan S. Abel)View Large ImageDownload PowerPointIn addition to the balloon pops, Icons of Sound has conducted further acoustic measurements in Hagia Sophia. In 2014–15 Turgut Ercetin recorded the acoustics of Hagia Sophia using sine sweeps driving loudspeakers as well as balloon pops. He used a pair of omnidirectional microphones and one A-format microphone capable of recording the spatial features of the sound field. One of the loudspeakers was placed next to the opus sectile mosaic on the floor of the nave. The three microphones were distributed on the axis aligned with the apse. Another set of measurements was conducted in the narthex, where the loudspeakers were set in the central bay aligned with the imperial gates and the microphones were distributed in the central bay as well as the second to last bay on the north side.Auralizations: Recovering Hagia Sophia's VoiceConvolution is a process for producing artificial reverberation. It is an algorithm that imprints the impulse response of the space being modeled onto an input audio signal, which can be a recording or live sound. Our convolution of the anechoic recording of Buhler's singing with the acoustic signature of Stanford University's Memorial Church, using an impulse response derived from a balloon pop, serves as a proof of concept.20Convolution done as a postproduction process, that is, using prerecorded sound to convolve with a room signature, does not allow musicians to interact with the space in real time.21 To overcome this limitation, we developed a real-time method for imprinting Hagia Sophia's sonic signature onto a live performance. This system makes it possible for musicians to hear the effect of the space on their singing in real time and make adjustments accordingly. They thereby can have substantially the same aural experience as performing in the real space.The balloon pop recorded by Pentcheva in Hagia Sophia, December 2010 (Fig. 3), was converted into an impulse response suitable for live auralization (Fig. 4).22 This recovered impulse response reveals the extraordinary acoustics of Hagia Sophia; it takes almost 12 seconds for the impulse response to decay from a comfortable listening level to the threshold of human audibility (approximately 60dB below conversation level). In addition, the listening position for this impulse response, about 30 m from the balloon, is dominated by reflected energy, or the "wet" portion of the impulse response. We can associate certain features of the impulse response with the geometric features of Hagia Sophia. For instance, the large peak around 100 ms (milliseconds) and the "wash" that builds to a wide peak around 300 ms issue from complex reflections produced by the dome and colonnades.In advance of the public concert, an experimental recording session was conducted in 2011 in a small recital hall (the Stage) at Stanford University's Center for Computer Research in Music and Acoustics (CCRMA). It featured thirteen members of Cappella Romana, a renowned group specializing in performances of Byzantine chant. To enable these professional singers to adjust aspects of their performance such as vocal balance and articulations in response to the reconstructed Hagia Sophia acoustics, the auralization system needed to operate in real time and to allow the chanters to hear each other and themselves in the simulated space.In this recording, we used Countryman Associates B2D headset microphones in order to capture each singer's voice on a separate track. A digital audio workstation (MOTU Digital Performer) was used to record the singers' voices, and we processed them to form a dry stereo mix. The mix was convolved in real time with stereo left and right statistically independent impulse responses derived from the December 2010 Hagia Sophia balloon-pop recordings, forming a wet stereo mix.23The chanters were positioned in a circle so that they could see each other and the conductor. Earbud headphones made available to them the dry and wet mixes, and each performer could control the balance between wet and dry mixes and the overall level in the individual's earbud signal. In this way, the performers could hear their own singing and each other in the simulated Hagia Sophia and interact accordingly. Providing feedback over headphones made it possible to keep the microphone signals dry, absent of any Hagia Sophia acoustics.The group began warming up without simulated acoustics; during this time microphone levels and equalizations were set. The simulated acoustics were then enabled, placing the performers in a virtual Hagia Sophia. Next, the performers were given a chance to acclimate to the environment. The chanters reported interacting reasonably naturally with the space, slowing their tempo to accommodate Hagia Sophia's long reverberation time. The ison (the drones) particularly enjoyed singing in the virtual acoustics because they found it easy to "ride" the resonances.The group performed a number of chants of the cathedral rite, including (1) a prokeimenon (psalm verses prefacing the scriptural readings, similar to the Latin gradual); (2) a kontakion, or sung sermon; and (3) a congregational setting for the fixed Psalm 140 (141) sung at vespers, all of which we processed in postproduction into stereo and surround recordings. Recording the individual tracks anechoically made a number of editing options available that would be precluded when recording such a group in the actual space using room mikes, as is conventional. This is because with room mikes the dry singing and acoustics of the space are intermixed. With the convolution applied artificially after the fact, the acoustics of the space in which the dry tracks were recorded will still appear in the final auralization. By contrast, close-miking allowed us to eliminate the acoustics of the recording room as well as stray noises, such as traffic sounds from outside the building.In 2013 Icons of Sound developed a method for live auralization of Byzantine chant for performance and recording in which we employed loudspeakers rather than headphones to synthesize the acoustics of Hagia Sophia. The use of loudspeakers both enables presentation to an audience and provides for a more natural interaction among the performers, allowing for freedom of movement and a truly shared performance environment. The use of loudspeakers also presents some technical issues, and in the following we discuss these challenges and outline our solutions.Our technical approach was driven by the need to perform, to rehearse, and to record in different spaces that were not preconfigured to present virtual acoustics and to which we had limited access. This required a virtual acoustics system that was transportable and could be quickly loaded in, installed, configured, and tuned; and also quickly torn down and loaded out. The system hardware is much like that of the headphone-based system described above, again including a set of close microphones to capture the dry voices of individual singers for recording and postproduction. A second set of room microphones is set to record the mix as heard in the space. The system also includes a set of powered full-range loudspeakers and subwoofers to play the simulated acoustics. A digital audio workstation (DAW) is connected to the microphones and loudspeakers through a mixing board and audio interface. The DAW processed the close-microphone signals to generate live virtual acoustics signals, rendered in the space by the loudspeakers.The use of loudspeakers raises the issue of feedback: it is possible, particularly when simulating very reverberant spaces such as Hagia Sophia, that loudspeaker signals will find their way back into the performer microphones, forming a feedback loop. This feedback can take a mild form, in which the simulated acoustics is modified, or it can take a severe form, in which particular resonant frequencies grow in amplitude and become unpleasant "whistles."The approach we take to minimize the possibility of feedback has two components. The first is to use many loudspeakers so that each loudspeaker can play a relatively quiet signal, thus minimizing feedback between any given loudspeaker and microphone. The difficulty is that there are a number of microphone-loudspeaker loops running in parallel, and their combination might create feedback if they operate coherently. This possibility is eliminated in our system by using statistically independent impulse responses—in effect, impulse responses that sound the same but do not track each other in any predictable way—to generate the simulated acoustics. When this approach is employed, a significant amount of reverberation can be generated without producing perceivable feedback.The second component that helps to eliminate feedback employs directional close microphones. Using microphones with a hypercardioid polar pattern (which is sensitive to sound coming from one particular direction) taped to the foreheads of the singers and pointed down toward their mouths (Fig. 7) places the singers' voices well within the "main lobes" of the microphones, and therefore they will be accentuated. At the same time, the loudspeaker signals, appearing from the sides and above the singers, will arrive from outside the main lobes of the microphones and therefore will be suppressed. We have found that this approach is sufficient to eliminate problematic feedback, even when simulating very reverberant environments such as Hagia Sophia and operating in smaller rooms, such as the CCRMA's Stage, where the loudspeakers are close to the singers. Two settings were configured for virtual acoustics performance and one for rehearsal and recording. The Bing Concert Hall was configured to simulate the acoustics of Hagia Sophia for the "Constantinople" portion of Cappella Romana's February 1, 2013, concert, From Constantinople to California.Fig. 7. Alexander Lingas, the artistic director of Cappella Romana, being outfitted with a lavalier microphone (Countryman B6, 2-mm diameter) by Scott Levine for the concert "From California to Constantinople," February 2013, at the Bing Concert Hall, Stanford. (© Dave Kerr)View Large ImageDownload PowerPointThe performance was held in the Bing Concert Hall just two weeks after the hall opened in January 2013. Here we devised a virtual Hagia Sophia using twenty-four full-range loudspeakers and six subwoofers.24 Twelve of the loudspeakers were arranged around the perimeter of the hall and twelve were hung from the ceiling, forming a "dome" of loudspeakers (Figs. 8–9). The fifteen performers, twelve melody chanters and three ison chanters, were outfitted with Countryman Associates B2D microphones and radio transmitters to provide dry chanter signals (Fig. 7). For recording, there were eight room mikes manufactured by DPA Microphones, two on stage and six flown in the hall.Fig. 8. Bing Concert Hall, Stanford University, live auralization loudspeaker layout for the performance "From Constantinople to California" February 2013 (diagram by Fernando Lopez-Lezcano)View Large ImageDownload PowerPointFig. 9. Cappella Romana at a rehearsal for the concert "From Constantinople to California" at Bing Concert Hall (© Dave Kerr)View Large ImageDownload PowerPointThe Bing Concert Hall is a rather reverberant space, having a 2.5-second-long reverberation time when its dampening curtains are deployed. Accordingly, the auralizing impulse responses were adjusted so that the resulting acoustics—the actual acoustics of the Bing combined with the virtual acoustics presented over the loudspeakers in the concert hall—produced a faithful rendering of Hagia Sophia. A set of Hagia Sophia impulse responses was imprinted on each dry signal, forming a "cluster" of reverberated signals for each performer. These clusters were then placed about the venue to create a set of overlapping regions in space for the performers' reverberated singing. Specifically, the virtual acoustics were generated by separately processing each of the twelve melody chanters' dry microphone signals. Four statistically independent Bing-corrected Hagia Sophia impulse responses were employed to form a "wet" signal cluster for each melody chanter. These clusters were then sized and positioned about the hall according to the singers' onstage arrangement using Ambisonics.25 Each of the three ison chanters was associated with a group of four melody chanters. Their singing was auralized by simply mixing their dry microphone signals with that of the four associated melody chanters prior to convolution and Ambisonics processing. In this way, their bass contribution would occupy a large region in space, as tends to happen with low-frequency sounds. The result was a reverberant sound field placed about and above the audience, with the men and women of the group layered about the dome of loudspeakers.Cappella Romana performers and several composers attending the concert reported that the space sounded and reacted in a very natural way; they found it easy to forget that the acoustics were simulated. A number of Cappella Romana members volunteered that the virtual Hagia Sophia sounded and responded like a real space, in contrast with their prior experiences with artificial reverberation. Many group members also commented that the virtual Hagia Sophia "held its pitch." The ison chanters thoroughly enjoyed the responsiveness of the simulated space to high-frequency harmonics and the ease with which the drone could be built and sustained (view video of a prokeimenon performed at Bing Hall.The Music and Choirs of The Great ChurchIcons of Sound's live auralizations have enabled us to focus some of our research on the interaction between the melodic compositions and the acoustics of the space. The study of the Constantinopolitan cathedral chant is a relatively recent phenomenon.26 Some of the ekklēsiastēs chants have survived in the later music collections of the Asmatikon (designed for the elite choir) and the Psaltikon (intended for the soloist).27 This sung office extends beyond the psalmody to include elite solo parts in the singing of the hymns known as Trisagion (at the Entrance with the Gospel book) and Cheroubikon (at the procession with the Eucharist gifts) as well as specific prelude pieces, such as prokeimena prefacing the readings from the Epistle in the Divine Liturgy (Eucharist rite) or introducing the Old Testament readings at hesperinos (vespers) or the New Testament at orthros (morning office); allēlouïai, before the recitation from the Gospels; koinōnika, or communion verses; hypakoai, identifying short monostrophic hymns similar to Western responsories; and finally, kontakia, or sung sermons, a genre created in the sixth century.28We have observed how the reverberant acoustics of Hagia Sophia challenge the intelligibility of both the spoken and sung word. The medieval music composed for the cathedral rite exhibits features that further increase this blurring of semantics. Melismas (singing many notes to a syllable) and intercalations (the inclusion of nonsematic, or "asmatic," sounds in between the syllables of a given word) obfuscate the intelligibility of the chanted word.We use the following example from the Psaltikon manuscript (Florence, Biblioteca Laurenziana, Ashburnhamensis MS Gr. 64, fol. 260v) to illustrate both the typical cathedral rite's melismas and intercalations, originally composed for the soloist (domestikos) and the elite choir of the psaltai.29 While this particular manuscript was produced in and served the liturgy of southern Italian monasteries, these institutions performed the cathedral chant of Constantinople on major feast days. Since this example records the last antiphon for the kneeling vespers on Pentecost, it offers reliable evidence from which to extrapolate aspects of the melody of the chanted office in the Great Church.30 Fol. 260v features Psalms 18(19).1 (Fig. 10 and the Auralization of this Teleutaion Antiphon). The phrase "glory of God" is rendered with an elaborate melisma; it unfolds across four lines of neumatic notations, showing the many musical tones stretching the chanted vocalization of these two short words. Melismas can produce rich prosody (sonic texture) at the expense of the semantics of words.Fig. 10. Florence, Biblioteca Laurenziana, MS Gr. Ashburnhamensis 64, fol. 260v dated to 1289, showing melisma on doxan theou and intercalation in the singing of variant B of the Allēlouïa refrain of the teleutaion antiphon for Pentecost vespers. (Reproduced with permission of MiBACT. Further reproduction by any means is prohibited.)View Large ImageDownload PowerPointThis blurring effect is only enhanced by the singing of the next word, allēlouïa, where we can witness how the second characteristic feature of cathedral chant—intercalation—functions. Non-semantic syllables break or stretch the semantic chains. We have marked them in bold for clarity's sake:Ἀχαουαχαουαλλεχεουεγγεενανενενεελουνουϊαγγα (Auralization of the Teleutaion Antiphon, Audio Animation 1)The intercalation of the nonsemantic kha-ou-a-kha-ou-a after the initial alpha challenges the semantics of allēlouïa; it breaks the natural syntagmatic unfolding of the word's vocalization in time, subduing semantics to prosody. The use of melismas and intercalations in this example for vespers on Pentecost manifests how the cathedral chant could produce performances at the limits of the human register of intelligible vocalization, pushing into a domain of sound that is audible but not always semantically accessible. The blurring of meaning with melismas and intercalations must have produced a sensually rich sound field, defying the rational, linear order of human speech.In his sixth-century ekphrasis Paul the Silentiary wrote how the chant of the elite choir singing from the ambo mixed the human voice (phōnē) with the divine utterance (omphē), making Christ aurally present like an icon of sound:31 "In its center the God-fearing hymnists (hymnopoloi) of Christ, by whose voice of immaculate breath/Spirit, the [divine] sound (omphē) that proclaimed the human childbirth of Christ came among men."32 Omphē is a Homeric word; it designates the voice that is beyond the human register of speech, metaphysical and incomprehensible.33 It is possible that the cathedral use of melismas and intercalations aimed to imitate this divine omphē, producing aural experience at the limits of intelligibility.The same passage from Paul the Silentiary further calls attention to the elite choir, describing them as hymnopoloi. Poloi can be translated both as
Referência(s)