Solve one room for reverberation, equipment noise and partition performance.
Three closed forms, each stated so a reader can check the arithmetic:
The Sabine absorption A = sum(S * alpha) and T60 = 0.161 * V / A;
The room constant R = A / (1 - alpha_mean), and at each receiver Lp = 10 * log10(sum over sources of 10^(Lw/10) * (Q / (4*pi*r^2) + 4/R)), the direct
field plus the diffuse field;
The composite transmission loss R_c = -10 * log10(sum(tau * S) / sum(S))
with tau = 10^(-R/10), which is why one weak panel dominates a strong
wall.
The exclusions are stated as loudly as the results. A single broadband
absorption coefficient cannot produce a speech transmission index, which
needs a modulation transfer function over an impulse response, so that metric
is unsupported rather than estimated.
The reverberation time, the room constant and the receiver level are all
statements about a diffuse field, so all three gap wherever there is none: a
room that absorbs nothing rings forever, and a room that absorbs everything
never rings at all. Neither extreme is reported as a number. The second is
the one worth naming, because Sabine does not diverge there but keeps
returning 0.161 * V / A, a positive decay time for an anechoic chamber,
which is the arithmetic of the model rather than a property of the room.
A receiver whose energy underflows to zero or overflows to infinity in
double precision has no level, so the level metrics gap with a reason naming
which bound was crossed.
Throws on a malformed request (see IAutoMovieAcousticRequest), on a
receiver within 1e-12 m of a source, and when the sealed run would fail
validation; otherwise the run is always solved, with one sound-pressure
sample per receiver when a level exists.
Solve one room for reverberation, equipment noise and partition performance.
Three closed forms, each stated so a reader can check the arithmetic:
A = sum(S * alpha)andT60 = 0.161 * V / A;R = A / (1 - alpha_mean), and at each receiverLp = 10 * log10(sum over sources of 10^(Lw/10) * (Q / (4*pi*r^2) + 4/R)), the direct field plus the diffuse field;R_c = -10 * log10(sum(tau * S) / sum(S))withtau = 10^(-R/10), which is why one weak panel dominates a strong wall.The exclusions are stated as loudly as the results. A single broadband absorption coefficient cannot produce a speech transmission index, which needs a modulation transfer function over an impulse response, so that metric is
unsupportedrather than estimated.The reverberation time, the room constant and the receiver level are all statements about a diffuse field, so all three gap wherever there is none: a room that absorbs nothing rings forever, and a room that absorbs everything never rings at all. Neither extreme is reported as a number. The second is the one worth naming, because Sabine does not diverge there but keeps returning
0.161 * V / A, a positive decay time for an anechoic chamber, which is the arithmetic of the model rather than a property of the room.A receiver whose energy underflows to zero or overflows to infinity in double precision has no level, so the level metrics gap with a reason naming which bound was crossed.
Throws on a malformed request (see IAutoMovieAcousticRequest), on a receiver within
1e-12m of a source, and when the sealed run would fail validation; otherwise the run is alwayssolved, with one sound-pressure sample per receiver when a level exists.