Embedded Files
CATT-Acoustic model of a shoebox-shaped room
Task 3
We varied the absorption so that five different reverberation time values in the range 0.3 - 2 s at 1kHz were obtained. In order to compare with the graph below (ATA figure 5.23) we decided to have absorbers on the walls (not the audience) so that the reverberation time is equal to 0.3, 1.2, 1.5, 1.7 and 2 seconds. The latter was found by using the absorbers on the exemple file from CATT-Acoustic. Respectively, the absorption coefficient at 1 kHz were 0.9, 0.2, 0.15, 0.12 and 0.07. Sabine's formula was used to find these values.
In order to understand the influence of the room we auralized the room with 3 different sounds (the same sound as in the previous tasks), i.e we took the impulse reponse of the room for each different configuration and convolved it with the speech, drums and orchestra music. The source is close to the front wall (on the drawing) approximately in the middle (on the x direction). In the first tasks the receiver is close to the source.
Reverberation time
It is quite clear that the files sound different from one "room" to another. The reverberation time of 2 seconds and the 0.3 seconds are the two edges of our study here, and one can see that the perceived reverberation time is very close to the real one. For instance the 0.3 seconds room seems to be typical of an anechoic room because one can't "feel" the reflections. The speech is very good for that kind of feeling because we are used to hearing speeches, but the drums can also be of use because one can hear the beats, dry and clear.
Comparison with the figure 5.23
The aim here is to compare the reverberation time of one sound and the perceived reverberation time depending on the value of H. H is a logarithmic ratio between the direct part of the acoustic energy and the reverberant one. Thanks to the figure 5.23 one can say that a higher value of H (meaning the closer to the source) gives a lower perceived reverberation time value.
In order to know in which field the receiver is (i.e reverberant or direct), we used this formula :
where r is the distance source-receiver. If the first term of the log is dominant then the receiver is in the direct field, and inversely. Then we could check the value of:
Receivers
Receiver 1 coordinates: (0;2;2)
Distance from the source:1.5m
direct field -> H= 6.5dB
Receiver 2 (5;10;1)
r = 9.8 m
reverberant field -> H= -10 dB
Receiver 3 (-3;30;3)
r =29.3 m
reverberant field -> H= -19 dB
We had 3 different receiver positions: the first one is close to the source (so expected to be ine the direct field); the second one is a bit further from the source; and the last one is far from the source, on the other part of the room (the receiver positions are shown in the shoebox picture above, one has just to to touch the screen with his nose to see them).
The reverberation time used is 1.7s (the middle curve on the graph 5.23). If one listens to the near field speech, one can notice that the perceived reverberation time is very short, shorter than 1.7s, and approximatly in the order of 1 second, which agrees the theory.
The middle distance speech has a longer perceived reverberation time, which again agrees the graph. The far distance has a value of H which is out of the scale of the graph, but one can clearly see the long perceived reverberation time.
Perceived reverberation time depending on the anechoic sound track
The orchestra is a sound very difficult to separate from its reverberation. An orchestra music is made of a lot of sources with different reverberation times, and since we are not much used to listen to that kind of music (except for orchestral music-lovers), ours ears are not used to detect a change of reverberation time.
The drums are made of several different beats so the reverberation time is perceived as shorter than the one for the speech. The speech is something we can listen to everyday so we can easily detect a change in the tone or in the duration of one word. Therefore we would say that the longest perceived reverberation time is the one for the speech.
Matlab file (to calculate H)
Page updated
Google Sites
Report abuse