The ancient theatre of Taormina
This drawing needs scripting and WebGL 2.
How this was made
Two simplified acoustic models compare a Roman reconstruction hypothesis (A) with the ruin configuration as simulated (B). Rays carry sound from a stage voice to recorded listening positions; the drawing uses separate geometry.
The example passes Adriano Evangelisti’s Latin excerpt of Cicero through statistical simulated responses. Arrival times and energies come from the simulation; reflected phase and fine detail are generated. These mono impressions establish no theatre measurements, calibrated loudness or speech intelligibility. The model has not been checked against measurements made in the theatre; the numbers come from the ray records.
Oration recording: Christoph Böhm et al. (2019), CC BY-NC-SA 4.0; performed by Adriano Evangelisti. Derived files retain that licence.
Sounds you can play here
Besides the oration, five sounds can be played through the same simulated responses: three recordings, your own voice and a file of your own. Each is heard at the chosen seat, in one building and with one audience at a time, mono, through the responses simulated for that seat. Nothing here is measured or calibrated, and the levels are set by this site, not by the theatre.
Tragedy: the opening of a reading of Euripides, Bacchae 1–63, in Attic Greek by Ioannis Stratakis (Podium-Arts, YouTube): its first 55.7 s, from the simulated voice’s place on the stage. Chorus: a modern performance of a chorus from Euripides, Orestes, “Stasimon Chorus from Orestes (chromatic performance)” (Steven Baur, YouTube, 2016): 77.6 s, from the simulated voice’s place on the stage; its loud and quiet passages were brought closer together by this site. Gladiators: a gladiator-fight sound-effect mix (Sound Effect Database, YouTube), 30.1 s, played as one sound at stage centre; it is not a historical recording. All three recordings keep the acoustics and processing they carry, which are not known to be anechoic; the theatre’s simulated response is added to them. The Bacchae reading and the gladiator effect were chosen and bought by the project owner, who supplied the files; the Orestes chorus was supplied by the project owner.
Your voice stands where the simulated voice stands. Use headphones: through speakers the microphone can howl. Its level is set when the microphone starts (and lowered once if it gets too loud), one gain for both buildings and both audiences. Your file plays as one sound at stage centre, mono, up to 4 minutes; it is read and played in your browser. Stage centre is the voice’s own place on the stage, 0.1 m lower, so the mark and the distances shown hold for the gladiators and your file within 0.1 m.
For the three recordings and your file, the plate at a seat draws the computed level through one pass in both buildings: the sound through that seat’s simulated response, at the level it is played and before the output limiter, with the building in the picture drawn dark. The recordings’ rows were computed in advance from the same responses; your file’s are computed in your browser.
When you change building or audience at a seat, the sound hands over from one simulated response to the other; changing the audience may skip or repeat up to 0.11 s. While you travel between seats, at the whole theatre or with the theatre divided, it is silent. An output limiter keeps every sound under one ceiling; the sheet says “Output protection on” while it acts.
Full record
This page draws the ancient theatre of Taormina live in your browser, first as a line drawing and then in stone. If you are reading this and see no theatre, the drawing has not started: it needs scripting and WebGL 2.
The ancient theatre of TaorminaHear one voice in the theatre as reconstructed and as it stands today. A simulation. One voice in both buildings. A simulation.Outline; the stone follows.Found, in the simulation with empty seats: the voice's computed strength is lower in the ruin than in the reconstruction at many of the 24 positions, and its clarity for music higher at some.What the simulation foundSixteen of the 24 computed positions have a recording to hear.Dotted circles: seats the stone hides from this view.
Scroll for the rest.
A full audience is simulated only as sound absorption on all the seating; the figures drawn with it are a picture, not counted and not part of the simulation.
Cuts instead of glides.
The simplified geometry and materials the computation used, in place of this drawing.
This drawing follows the published surveys and reconstructions; it is not the computation’s own geometry. Seats are placed by their row and section. Sound and results use separate simulation models A and B.
The phrase repeats. Simulated sound, one channel, not a calibrated loudness.
Precision, in full
Scroll for the rest.