Impulse Response
Reflection analysis, surface inference, RT60 estimation, and room mode identification from acoustic impulse responses
Reading the Impulse Response
An impulse response (IR) shows how a room responds to a single, instantaneous burst of sound. Every peak in the IR represents a sound arrival — the direct sound followed by reflections from surfaces in the room.
AudioMap captures impulse responses using ESS (Exponential Sine Sweep) or MLS test signals and extracts the IR through deconvolution. The resulting time-domain waveform reveals the complete acoustic signature of the space.
Key Features
Direct Sound
The first arrival — the strongest peak in the IR. Represents the straight-line path from speaker to microphone.
Early Reflections
Arrivals within the first 50-80ms. First-order reflections from walls, floor, and ceiling.
Late Reflections
Dense reverberant tail after 80ms. Characterizes the room's reverberation.
Noise Floor
The background noise level visible as the IR decays. Limits the usable dynamic range.
Reflection Detection & Analysis
AudioMap automatically identifies individual reflections in the impulse response, estimating their timing, level, and originating surface.
Peak Detection
Identifies significant peaks in the IR above a configurable threshold. Each peak represents a discrete reflection arrival.
Time-of-Arrival
Calculates the delay of each reflection relative to the direct sound. Combined with the speed of sound, this gives the extra path length traveled.
Level Analysis
Measures the amplitude of each reflection in dB relative to the direct sound. Helps assess which reflections are perceptually significant.
Surface Inference
AudioMap's surface inference engine uses a Generalized Hough Transform combined with Expectation-Maximization (EM) refinement to identify which room surfaces are responsible for each detected reflection. The system works with arbitrary surface normals and room geometries.
Processing Pipeline
Hough Voting
Each reflection casts votes in parameter space for possible surface locations
EM Refinement
Iteratively refines surface parameters using probabilistic assignment
Model Selection
Chooses the best surface model using information criteria
Candidate Surfaces
Builds final surface candidates with confidence scores
Post-Processing
Validates results and assigns reflections to surfaces
3D Visualization
Detected surfaces are rendered in AudioMap's 3D room view, showing the reflection paths from speaker to surface to microphone. This provides an intuitive visual understanding of the room's acoustic behavior.
RT60 Estimation
RT60 (Reverberation Time) measures how long it takes for sound to decay by 60 dB after the source stops. AudioMap calculates RT60 from the impulse response energy decay curve (Schroeder integration).
T20
Extrapolated from -5 dB to -25 dB decay range
T30
Extrapolated from -5 dB to -35 dB decay range
EDT
Early Decay Time — from 0 dB to -10 dB, reflects perceived reverberance
Room Mode Analysis
Room modes are standing wave resonances that occur at frequencies determined by the room dimensions. AudioMap identifies room modes from the impulse response and frequency response data.
Axial Modes
Between two parallel surfaces. Strongest and most problematic. f = c/(2L)
Tangential Modes
Involving four surfaces. Moderate energy, ~3 dB weaker than axial.
Oblique Modes
Involving all six surfaces. Weakest but most numerous.
Room modes are most significant below the Schroeder frequency, where individual modes are well-separated and dominate the room response.