alexanderlerch/ACA-Code

double check references

alexanderlerch opened this issue · 1 comments

mel = 1000 * log2(1 + f/1000);

Should we use the linear scale below 1000Hz? Also impacts ToolMel2Freq (and revisiting the python code might be necessary, too).

It seems that there is no linear part for the Fant formula. That apparently stems from Koenig in a paper I do not have access to: Koenig, W, A New Frequency Scale for Acoustic Measurements. Bell Laboratory Report 27, pp. 299-301, 1949.