Integrated impulse response processing

This menu handles the parameters used to calculate reverberation times according to ISO 3382. The procedure is based on the analysis of the reverse-time integrated impulse response (IIR).

The following plot shows a two-channel impulse response measurement, recorded in a reverberation chamber using a dodecahedron loudspeaker and a sweep signal for excitation.

For any further evaluation at first the segment of the impulse response to be considered in the processing needs to be determined. For this the respective region is selected under Select region. The endpoint for the evaluation should be chosen carefully. The impulse has to merge neatly into the noise floor, in order to enable an automatic detection of that very noise floor. Furthermore the noise floor should not be polluted by distortions in the signal chain, especially from the loudspeaker. When using sweeps for measurements, the distortion artefacts appear at the end of the impulse response. In the plot above this is found in the rising slope setting in at around 16s.

To be on the safe side, also the impulse start (see next plot) should be examined more closely when setting the starting point in order to identify disturbances that could affect the automatic detection of the impulse start. In that case the signal might have to be edited or an adapted starting point would have to be chosen. For a safe detection of the impulse start its dynamic range has to be at least 25dB. If there are major disturbances present before the impulse start and if through failure of Start detect they are included in the integration, then especially the results for EDT and T10 might be falsified.

Under Processing options the following options are to be configured:

  • IR start detect: use of automatic detection of impulse response start.
  • Thr. rel max [dB]: defines the impulse response start relative to the maximum of the impulse response in dB for IR start detect; if the threshold is set to -20dB, the impulse response start is being found at the point in time before the actual impulse peak, where the response exceeds the global impulse maximum minus 20dB.
  • Scan band: if set to each, after filtering the impulse response, for each frequency band the IR start detection is carried out and used for further processing; if set to highest only, only for the highest frequency band the detection is carried out.
  • IR end detect: automatically identifies the point in time, where the filtered IR merges into the noise floor.
  • Tailcut compensation: compensates the error caused by the finite integration time at the end of the IIR (dropping of IIR close to the noise floor) according to ISO 3382-1, chapter 5.3.3.

Noise subtraction contains several parameters to be specified for estimating the RMS-value of the noise. This is subtracted from the impulse response before the integration in order to increase the accuracy for the calculation of decay times.

  • Noise subtraction: if activated, the subtraction of the noise voltage is being applied.
  • Noise detect: determines the noise voltage from the region defined by the parameters Take RMS from and Noise length. If Noise detect is deactivated a value for the noise voltage can be entered manually in Noise RMS.
  • Take RMS from: head examines the noise in the region before the impulse start, tail looks behind the transition region after the impulse response; lowest segment finds the region with the lowest noise level.
  • Noise length [%]: defines the length of the noise floor segment to be evaluated as a percentage of the respective region. E.g. if the noise tail is 10s long, and Noise length is set to 10%, then the first second of the tail is evaluated to obtain the RMS-value of the noise.

Filter Options allow Fractional octave filtering of the impulse responses according to DIN EN 61260.

After starting the calculation by OK, the display switches to Int. impulse response domain. Here the actual Decay time analysis can be performed. The following plot shows IIRs of a two-channel measurement evaluated in octaves.

If errors occur during the calculations, for example when the noise start could not be identified due to insufficient SNR, a window appears listing the errors. Depending on the kind of indication, processing errors can be avoided by suitable adjustments of starting and end points for the region to be evaluated (Select region).