Interactive / Room acoustics
Hear the room.
Same voice. Different conditions.
Change the surfaces, the competing sounds, where you listen and how the room fills.
Room acoustics is one part of a sound experience. For music choice, emotion and autonomic responses, explore music & psychology.
These are educational examples of acoustic relationships, rather than a calibrated model of a particular café, restaurant or product. Try headphones for the stereo differences.
Press Play to listen.
Change the room.
Start with hard surfaces. Add absorption and listen to what lingers after each voice.
Press Play to follow the sound.
Speech-linked wavefronts travel to the walls, reflect and reach the listener. Motion is slowed to make the paths visible.
Choose the surfaces
Hard floor, walls and ceiling allow more sound to linger.
A longer tail after the sound stops.
Change a setting, then replay. The source recording and volume setting stay the same.
Absorption changes how reflected sound builds up and decays. It does not remove the voice itself or soundproof the room.
Across the table.
Listen to a voice, or two people taking turns across a table. Change the café music, other conversations and room surfaces.
Press Play to follow the sound.
For one voice, you sit across the table. For two voices, you listen nearby while the female and male speakers take turns. Background voices stay at fixed positions.
Relative mix controls, not measured sound levels. Turn both down to hear the voice alone in the selected room.
A shorter room tail can help a voice stand out, but background sound still matters. Listen for the different roles of room reflections and competing sound.
What is being compared?
Single-voice playback puts you 1.5 m across the table. With two voices, the speakers stay across from each other and your listening position moves beside the table. Two background voices and an original café instrumental form adjustable layers. Absorption changes the room response for each source.
This example does not model diners raising their voices as the room gets busier, or calculate a speech-intelligibility score.
Follow the sound.
Move the voice and the listener. Trace the direct route and first wall reflections.
Press Play to follow the sound.
Drag either marker, use its arrow keys, or adjust the position controls. The listener faces the top of the plan.
Listen for changes in distance, stereo direction and the reflected tail. Stereo is illustrative, without head tracking.
Explore the reflection routes
Only the direct and first wall-reflection routes are drawn. The audio also includes first floor and ceiling reflections and a synthetic late tail.
As the room fills.
More people absorb some sound. More conversations add sound. Separate the two effects.
Press Play to follow the sound.
Counts include the two people in the foreground. The other markers show occupancy and speaking activity; they do not act as individually traced obstacles.
The foreground recording stays fixed.
Reference: 12 seated diners, 4 other talkers, hard surfaces, 0.35 m² per person and voice feedback on. The dB difference is modelled; your device volume is not calibrated.
Try two different questions.
Keep other talkers at four and add seated people. The model adds absorption and shortens the tail. Then keep occupancy fixed and increase the number speaking. Compare the rise in competing sound.
The feedback switch uses Rindel's 0.5 dB/dB Lombard relationship above its 45 dB noise onset. It changes the level of the background voices, rather than synthesising the vocal effort or behaviour of real diners.
Grounded in Rindel's model.
“the group size g should not be considered constant, but varies according to the social character of the gathering.”
J. H. Rindel (2010), p. 1158. Read the author's paper.
The diffuse-field equations inform the relative background level and added seated-person absorption. This small room has not been validated against measurements; the original comparisons involved larger food courts and canteens. Low ceilings and strong absorption can weaken the diffuse-field assumption.