annotate Code/Descriptors/yin/private/src/dftoperiod2.m @ 4:92ca03a8fa99
tip
Update to ICASSP 2013 benchmark
| author |
Dawn Black |
| date |
Wed, 13 Feb 2013 11:02:39 +0000 |
| parents |
ea0c737c6323 |
| children |
|
| rev |
line source |
|
dawn@0
|
1 % prd=dftoperiod2(d,b,t) - estimate period from difference function
|
|
dawn@0
|
2 %
|
|
dawn@0
|
3 % prd: row matrix of period estimates
|
|
dawn@0
|
4 %
|
|
dawn@0
|
5 % d: column vector or matrix of difference functions
|
|
dawn@0
|
6 % b: bounds matrix ([lo, hi])
|
|
dawn@0
|
7 % t: threshold
|
|
dawn@0
|
8 %
|
|
dawn@0
|
9 % Difference functions are supposed to be cumulative-mean-normalized.
|
|
dawn@0
|
10 % For each column of d, search for the first minimum between lo and hi that
|
|
dawn@0
|
11 % falls below threshold. The index of this minimum (re 0)
|
|
dawn@0
|
12 % is the period estimate.
|
|
dawn@0
|
13 %
|
|
dawn@0
|
14 % This version differs from dftoperiod in that the threshold is
|
|
dawn@0
|
15 % incremented by the global minimum of the difference function.
|
|
dawn@0
|
16 %
|
|
dawn@0
|
17 % Mex function.
|
