Reply
A canyon seen from above, four hundred meters across, and you are the small bright point. Click anywhere to stand there and clap. The wave leaves you at the true speed of sound, each cliff returns it after exactly the flight time there and back, and the moment an echo reaches your ear a circle blooms at the measured radius, tangent to the cliff that answered. Drag the cliff ends to reshape the canyon; the echoes move with the walls. Yesterday's piece showed what sound cannot tell you about shape. This is what it can.
the room measured by ear
The left number is computed from the drawn geometry; the right is read blind from the trace, as arrival time times the speed of sound over two. Arcs beyond the walls are real too: corners answer, and two facing cliffs answer twice.
Yesterday, Hear staged the Gordon, Webb and Wolpert drums: two shapes no ear can tell apart, because the spectrum forgets the shape. But stand in a canyon and clap, and sound tells you exactly where you are. The arrival time of each echo is twice a distance over the speed of sound, and nothing about that is ambiguous. The eigenvalues cannot say which drum; the arrival times say precisely which wall, and where. The same operator, listened to in time instead of frequency, keeps what the spectrum throws away.
The wave here is not an animation of an echo. The acoustic wave equation is integrated live on a staggered grid, the scheme whose energy is conserved to fourteen digits and whose reciprocity holds to machine precision on the gate board. The ear's trace is the pressure at your cell, with your own clap subtracted the way your ear ignores your own voice. The detector that places the arcs is the same code the gates ran blind: in two seeded canyons it had never seen, it located every wall to within a meter and a half, and every voice it heard in the first-order window was accounted for.
What is tuned is admitted: the brightness of the field is auto-exposed with a floor, and the audible clap is a rendering. At canyon scale the honest waveform is infrasonic, so what you hear is the measurement played back, each detected arrival struck at its true delay and relative strength. The microphone mode runs your voice through the same measured delays and nothing else; frames of audio never leave the machine.
what the echoes cannot say
Two cliffs facing each other reply twice: the second bounce arrives as one more voice, drawing an arc tangent to no wall. On the gate board that phantom outranked a real cliff, and the honest resolution was not to filter it away but to name it: a room with opposed walls really does answer twice, and one ear alone cannot always tell a wall from a conversation between walls. The reconstruction is exact about distances and silent about directions. That is the price of listening from a single point, and the piece keeps the phantom arcs on screen because they are true recordings of a real reply.