Impulse Response Evaluation Set
explaination of metrics Matlab file.m Example of Schröder method
How to describe a room?
It would be very interesting to have simple numbers, coefficients or indicators in order to specify the quality of a room. The main problems to describe the room is that its quality is something subjective and that there are a lot of parameters which can influence our sensation of this quality (eg: the visual sensation of the room). Moreover if one wants to describe a room, one has to do it for a specific use (eg: speech, music...).
For all that, some metrics have been created, such as the reverberation time, the definition, the clarity, the room gain, the early decay time, the warmth and the brilliance... All those metrics characterise how sound is perceived in a room.
An evaluation set has been set up in order to describe the sensitive room properties using the impulse responses studied before. A brief explanation of the metrics used here to describe the investigated room can be found here.
Here are the average values of the explained metrics for each octave band. The average was made over all microphone positions. A filter has been used in order to have a single value which is the average of the whole octave band.
First of all one has to know that the Definition metric is more interesting than the Clarity since the latter is used to describe room for music.
The higher the frequency the closer the T60 and EDT. However the metrics differ between the two loudspeaker positions, and again it is more noticeable in low frequencies than in high frequencies.
metric values
According to the ISO 3382 our values of the reverberation time and the EDT are way too low. But those recommended values are set up for concert hall which actually need longer reverberation time than a room for simple speech as investigated here. Our values seem ten be reasonable since a long reverberation time can be very fast perceived as perturbation for the audibility of the speech.
The different values of the definition seem to be also reasonable since high values mean that the room is suitable for speeches. For most of the frequency bands and especially in high frequencies, most of the energy is in the first 50ms.
The warmth metric is used for music hall where classical music is played, it does not say a lot on our room. A recommended brilliance value is about 0.8-0.9. Here the values are lower. One can feel a lack of brilliance in that lecture room.
comparison with Room-Capture values
Frequency [Hz] 125 250 500 1k 2k 4k
EDT [s] 0.6 0.61 0.43 0.42 0.47 0.45
T30 [s] 0.53 0.52 0.49 0.46 0.43 0.44
C80 [dB] -2.6 0.9 6.5 7.1 6.5 7.5
D50 [%] 6 19.5 63.9 62.8 60.5 63.2
The reverberation times fit pretty well with the calculated ones. For the three other metric, higher the frequency colser the measured and calculated values. This might be due to the different filters used in the software or in matlab.