The Sound Book: The Science of the Sonic Wonders of the World by Trevor Cox Read Free Book Online Page B

Dempster and his fellow musicians as creating an “intensely serene music in which the slightest changes seem cataclysmic, and gradual swells emerge as tidal waves.” 51 Writing in the Times , Debra Craine describes the music as having an “eerie, majestic calm that envelops you with hypnotic elation.” 52 Notes played seconds apart form lush layers on top of each other, requiring a player to think about the interaction of notes played far apart; otherwise, intense dissonance is produced. Stuart Dempster commented, “Usually when you stop for a mistake, the mistake has the decency to stop too, but it doesn’t [in the cistern;] it just sits there and laughs at you . . . You have to be a clever composer [or improviser] and incorporate all your errors into the piece.” 53
    I listened to the album, enjoying the meditative polyphony, but also listening for the ends of phrases, because after the musicians stopped playing, the sound would naturally ring around the cistern. From these parts of the music, the reverberation time can be estimated. For over a decade, colleagues and I have been developing ways of extracting reverberation time from speech and music. The idea is to make measurements in concert halls, railway stations, and hospitals while they are in use. Conventional reverberation measurements require loud sounds: gunshots or loudspeakers blaring out noise or slow glissandos. These are unpleasant to listen to and can damage hearing. Audience members also have an annoying habit of ruining the results by commenting on the noise—“Wow, that was loud”—as the decay is being measured. But the sound of an orchestra in a concert hall, or the speech of a teacher in a classroom—while imperfect for measurement—does include the room acoustic; the difficult part is finding a way to extract the effects of the room from the music or speech. One of the most exciting areas of research at the moment is the use of computer algorithms to extract information from audio. A well-known example is Shazam, an app that identifies music from a brief recording through a mobile phone’s microphone. Other algorithms try to transcribe music automatically or identify the genre of unlabeled audio files.
    Applying our algorithm to Stuart Dempster’s recording gave an estimated reverberation time of 27 seconds over the low frequency ranges of the trombone and didgeridoo. 54 This is a good indication that the American cistern beats the Scottish reservoir. But to be sure, I wanted a conventional impulse response. When creating a new auditorium, acoustic engineers work from graphs and tables of reverberation times and other parameters to check that the hall meets design specifications. However, these scientific charts and parameters mean little to architects, so acousticians are increasingly making audio facsimiles of a proposed auditorium and getting clients to listen. This auralization starts with a piece of music that has been recorded in a completely dead space, like an anechoic chamber (described in Chapter 7). In other words, it is the sound of the orchestra without any room. Acousticians then combine this music with a model of how sound will move in the future place. In the past, impulse responses came from scale models of the auditorium at one-tenth or one-fiftieth of the full size, but nowadays they are more often predicted by computers.
    Auralization also works with impulse responses measured in real rooms, so it has also been encoded into artificial reverberation algorithms used by musicians and sound designers creating film and game soundtracks. In one of these reverberators I stumbled across a library of impulse responses that included three measured in the American cistern. At low frequency the Dan Harpole Cistern has the same reverberation time as the Wormit reservoir: 23.7 seconds. But at midfrequency the America cistern wins, with a reverberation time of 13.3 seconds. These