Archaeology meets DSP: CCRMA at Chavín de Huántar

In 2008, a team of researchers from CCRMA at Stanford University traveled with Professor John Rick to Chavín de Huántar in Peru. The purpose of the joint expedition was to measure and archive the unusual acoustics found on the site, as detailed in my previous blog post.  The team of investigators from CCRMA consisted of some notable luminaries from the fields of computer music and audio DSP, including Julius Smith (pioneer of waveguide synthesis), John Chowning (pioneer of FM synthesis), Perry Cook (expert in physical modeling), and Jonathan Abel (co-founder of Universal Audio, created many of the algorithms used in the UAD1 and UAD2).

The preliminary results from the expedition can be found at the website of the Chavín de Huántar Archaeological Acoustics Project. I highly recommend visiting the CCRMA site, and tracking down the various papers. A brief summary of the published results:

The CCRMA researchers brought a a novel microphone array to Chavín, to capture the impulse response of the galleries. The microphone array was specifically developed for room acoustics analysis and synthesis, and archaeological acoustics applications., and an overview of its design and construction was presented in an AES Conference paper, “A Configurable Microphone Array with Acoustically Transparent Omnidirectional Elements.” The new device consists of a number of omnidirectional microphone elements, mounted on flexible wire mounts, attached to a lightweight yet sturdy base that is suited for the narrow tunnels found at Chavín.

The microphone array is used in conjunction with a calibration system, which consists of 4 small speakers that are configured around the base of the microphone array:

By sending calibration signals through the speakers and processing the results, the various time differences between the microphone elements can be compensated for. The system is purported to obtain better results at capturing impulse responses of various structures than the previous microphone arrays that have been used.

The initial analysis of the acoustics at Chavín, recorded with a simpler microphone/monitor setup than described above, was published in a paper presented at Acoustics ’08 in Paris, “On the Acoustics of the Underground Galleries of Ancient Chavín de Huántar, Peru.” The researchers found that the reverberation times at Chavín were fairly short, on the order of 150 milliseconds to around 1 second. The paper suggests that the short reverb times would work for the rhythmic playing of the Strombus trumpets found on site. The reverb time increases as a function of the number of turns between the source and the receiver, with sources several gallery turns away from the receiver having a longer perceived reverb time.

The reverberation in the Chavín galleries is characterized by dense and energetic early reflections, and low inter-aural cross-correlation. All 3 of the galleries have a quick onset, where the reverberation reaches Gaussian statistics within 20 milliseconds of the initial impulse. The quick build to Gaussian (i.e. random) statistics, and the low amount of cross-correlation between the left and right ears, is responsible for the strange sonic characteristics of the galleries, where it is difficult to localize where a signal is coming from in the absence of a direct signal. David Griesinger, the pioneering DSP guru behind the original Lexicon algorithms in the 224/XL and the 480L,  has discussed the role of low inter-aural cross-correlation, or decorrelation, in creating a sense of envelopment, where the sound is perceived as surrounding the listener. In artificial reverberators, decorrelation is obtained by having different delay times or phases for the different output signals. At Chavín, the small distance between structural surfaces is probably responsible for the quick build of echo density to the late field, which demonstrates randomness down to the binaural level.

The acoustics research and analysis of Chavín is ongoing. There are a few issues which I feel would be interesting to address in future publications:

  • The galleries at Chavín were originally covered with plaster. It is possible that the original plaster would create a dramatic difference in the RT60 of the galleries. An example would be a small room made of wood, versus the same room covered in several coats of cement plaster – the latter describes the famous reverberation chambers at Gold Star Studios. The CCRMA website mentions that research will be conducted into the sonic effects of the plaster used at Chavín.
  • The analysis of the galleries at Chavín used a long swept exponential sinusoidal test signal for the impulse. Such a signal is useful in reducing the effects of noise on the analysis, but it also “smooths out” any time variation that might have been present in the original reverb response. It would be interesting to analyze the reverb of the galleries, to see if time variation would have any marked acoustic effects. The temperature of the galleries has been measured as fairly constant, but this is probably assuming modern sources of light, such as flashlights, as opposed to torches or the like. In addition, the presence of living bodies in a small space can have marked effects on the air temperature, which can result in small changes in the speed of sound that have a noticeable effect on the sidebands of a signal in the late reverb decay. The current reverberation time in the galleries is short enough that small variations in the speed of sound may not have a significant effect. However, the longer reverb times that may be associated with the original plastered walls may have allowed for time variation to be perceivable in the reverb decay, especially as heat sources are introduced into the gallery.

As a fitting end to this post, here is a video of Tita la Rosa playing a Strombus trumpet, presumably within one of the galleries at Chavín:


About the author:

Sean Costello is the "algorithmic reverb plugin wizard" [citation needed] at Valhalla DSP.

Comments (2)

  1. Your last point is a bit damning, and to me, undermines the pertinence of their findings. To wit, if one were to strip a home theater of the walls down to the concrete, one should expect to have a very different, and entirely irrelevant set of reverberation characteristics.

    To measure Chavín without the original environment (as you state, torches, bodies, plastered walls, also whatever pots and other baffles and reflectors, incidental or intended) is to measure something as irrelevant as the *present* audio characteristics of the ruins. It would be as impossible to extrapolate the original characteristics as it would be to model my theater echo characteristics after having stripped its surfaces.

    Now, to be fair, that’s exactly what I did, in order to identify major standing waves and nulls prior to building it out, but it could hardly be an accurate measurement of the reverberation characteristics of the final product.

    Also, I would like to note that their microphone looks very much like an Old One or a Mi-Go, aka Fungi from Yuggoth. As so much of those extraterrestrial senses relied on various forms of vibrational analysis, I will have to insist that you explore this in a future article.

    1. I agree with you to a certain extent on the effects of the plaster on the walls in the Chavín galleries. However, the envelopment or decorrelation should remain the same with the plaster present, as this is more a function of the shapes and size of the structures, as opposed to the Sabine coefficients of the wall coating (like how I just whipped out “Sabine coefficients”? It’s a term for the absorbtion/reflection of sound waves in a reverberant space).

      My guess is that the plaster would result in a longer T60 overall, but this is dependent on the type of material used. The frequency response may change as well. If the RT60 is increased, this would serve to further increase the sense of envelopment, as well as disorientation, as the ratio of reverberant to direct sound would increase. It would also affect the sound of the Strombus trumpets – a longer RT60 would result in the rhythmic articulations mentioned in the article being lost.

      A quicker summary: The plaster won’t change the reflections or modes in the space, just how long the modes hang around.

      Lovecraftian nature of the microphone array duly noted. In a future blog post, I will discuss recent studies of a staircase at Tikal, where the echo of a hand clap produces an impulse response that sounds out the syllables: “Iä! Iä! Shub-Niggurath!”

Leave a Comment

Rules of the Blog

Do not post violating content, tags like bold, italic and underline are allowed that means HTML can be used while commenting.