ISO1996-2:2007 Annex C
Tonal Adjustment Kt ISO 1996 (dB) is based on the Annex C of the ISO1996-2:2007 standard for the Objective method for assessing the audibility of tones in noise. The measurement procedures allow you to verify the presence of audible tones and, based on their prominence, recommended levels of adjustment.
The method is based on the concept of critical bands. The method outputs a measure of the prominence of the tones, the so-called "Tonal Audibility". An adjustment between 0 and 6 dB is applied to the sound level measurements in dB A.
The method's has three steps and they are presented below:
A-weighted narrow-band frequency analysis.
The frequency resolution Delta ƒ is set to 3.33 Hz so that the effective analysis bandwidth Beff (which is 1.5 times greater than the frequency resolution with a Hanning window) is equal to or smaller than 5% of the lowest critical bandwidth (100 Hz). This corresponds to a 300 ms signal window. Because this spectrum calculation will be further averaged over time, it is recommended by the standard that the signal duration be at least one minute,
Determination of the average sound pressure level of the tones and that of the masking noise within the critical band around the tone(s).
Tones are detected within the averaged spectrum according to an automatic procedure. First, "noise pauses" (groups of bins of the spectrum where the presence of tones can be suspected) are detected by comparisons of successive spectral bin levels over the frequency scale. Within each detected noise pause, a tone is detected if (1) the maximum spectral bin level is at least 6 dB greater than the first spectral bins outside that noise pause and (2) the 3-dB bandwidth around that spectral bin is smaller than 10 % of the critical bandwidth centered on it. If so, all spectral bins with levels within 6 dB of the maximum bin level constitute the tone, and its frequency is that of the maximum level.
The tone level Lpti is calculated as the energy sum of the level of the identified bins:
where Lt is the level of each spectral bin of the tone, Delta ƒ the frequency resolution, and Beff the effective analysis bandwidth.
The total tone level Lpt within a critical bandwidth corresponds to the energy summation of all included tones:
The noise level within a critical band is calculated by performing a linear regression through all spectral bins that were not classified as tone, over a range of +/- 75% of the critical bandwidth. The noise level Lpm is calculated on the basis of the energy sum of the values of the regression line Ln within the critical band:
Calculation of the Tonal Audibility Delta Lta and the adjustment Kt.
The Tonal Audibility Delta Lta within a critical band is calculated as follows:
This calculation is repeated for each critical band centered on a detected tone or a pair of detected tones ( centered at the mean frequency) provided these two tones belong to a common critical band and are the two most prominent ones of that critical band. The maximum Delta Lta value obtained defines the decisive critical band. Then the final adjustment Kt is calculated as:
Sound: Analysis and Specification makes it also possible to calculate Tonal Audibility and Adjustment according to the standard on a non-stationary basis. In this case, the signal is segmented into several time blocks, with a given overlap rate, and the calculation is performed in each block. This mode outputs a Tonal Audibility vector and an Adjustment vector over time. This mode is activated by ticking in the box corresponding to the Non stationary mode field in the parameter window of the tonality of ISO1996-2 Annex C. The window length and overlap rate can then be set to the desired values. Note however that too short a window length can alter the precision of the noise level estimation and as a consequence that of the computation results.