|
c@0
|
1 %% MASSEF.execute
|
|
c@0
|
2 % Run MASSEF using input mixtures and separators.
|
|
c@0
|
3 %
|
|
c@0
|
4 %% Syntax
|
|
c@0
|
5 %
|
|
c@0
|
6 % massef.execute(mixtures,separators)
|
|
c@0
|
7 % massef.execute(mixtures)
|
|
c@0
|
8 %
|
|
c@0
|
9 % |massef.execute(mixtures,separators)| runs the multichannel audio source
|
|
c@0
|
10 % separation evaluation framework (MASSEF) using the <help_MASSEF.html
|
|
c@14
|
11 % MASSEF> instance |massef|, the array of <matlab:doc('iosr.bss.mixture')
|
|
c@14
|
12 % |iosr.bss.mixture|> objects |mixtures| and the array (or cell array) of
|
|
c@14
|
13 % instances of separation algorithms contained in |separators|. Separation
|
|
c@14
|
14 % is performed for all combinations of separators and mixtures. The
|
|
c@14
|
15 % separation algorithm instances contained in |separators| should conform
|
|
c@14
|
16 % to the <help_Separators.html required format>.
|
|
c@0
|
17 %
|
|
c@0
|
18 % The |execute| method performs evaluations of the algorithm according to
|
|
c@0
|
19 % the data returned by the algorithm, and the options provided to
|
|
c@0
|
20 % <help_MASSEF.html MASSEF>.
|
|
c@0
|
21 %
|
|
c@0
|
22 % If the separation algorithm returns a signal, then |MASSEF.execute|
|
|
c@0
|
23 % evaluates:
|
|
c@0
|
24 %
|
|
c@0
|
25 % * signal-to-noise ratio (SNR);
|
|
c@0
|
26 % * signal-to-ideal-noise ratio (SINR) (the SNR with respect to the signal
|
|
c@0
|
27 % reconstructed using the ideal binary and ratio masks).
|
|
c@0
|
28 %
|
|
c@0
|
29 % In addition, if |evalPEASS| is true, then PEASS and BSSeval evaluation is
|
|
c@0
|
30 % performed. If |evalSTOI| is true, then STOI evaluation is performed.
|
|
c@0
|
31 %
|
|
c@0
|
32 % If the separation algorithm returns a mask, and if the ideal mask
|
|
c@0
|
33 % dimensions match the estimated mask dimensions, then |MASSEF.execute|
|
|
c@0
|
34 % evaluates:
|
|
c@0
|
35 %
|
|
c@0
|
36 % * ideal mask ratio (IMR) with respect to the ideal binary and ratio
|
|
c@0
|
37 % masks.
|
|
c@0
|
38 %
|
|
c@0
|
39 % If the separation algorithm returns a mask but no signals, then MASSEF
|
|
c@0
|
40 % attempts to calculate the estimate signal in order to calculate the
|
|
c@0
|
41 % signal-related metrics mentioned above.
|
|
c@0
|
42 %
|
|
c@0
|
43 % The |execute| method also calculates and evaluates the ideal binary and
|
|
c@0
|
44 % ratio masks using the signal-related metrics utilised for the algorithm
|
|
c@0
|
45 % evaluation. The masks are calculated using the settings that are local to
|
|
c@14
|
46 % each <matlab:doc('iosr.bss.mixture') |iosr.bss.mixture|> object.
|
|
c@0
|
47 %
|
|
c@0
|
48 % Lastly, the framework captures the estimated azimuth and elevation of the
|
|
c@0
|
49 % targets and interferer(s), if they are returned.
|
|
c@0
|
50 %
|
|
c@0
|
51 % Once the evaluation is complete, the <help_MASSEF.html MASSEF> object,
|
|
c@0
|
52 % which contains the results data in |MASSEF.results| as a
|
|
c@0
|
53 % <help_MASSEFresults.html MASSEFresults> object, is saved to the file
|
|
c@0
|
54 % given by |MASSEF.results_filename|. The separation and mixture objects
|
|
c@0
|
55 % are also saved.
|
|
c@0
|
56 %
|
|
c@0
|
57 % |massef.execute(mixtures)| evaluates only the ideal masks.
|
|
c@0
|
58 %
|
|
c@0
|
59 %% See also
|
|
c@0
|
60 % <help_MASSEF.html MASSEF>, <help_MASSEF_evaluate.html MASSEF.evaluate>,
|
|
c@0
|
61 % <help_MASSEFresults.html MASSEFresults>,
|
|
c@0
|
62 % <matlab:doc('iosr.bss.generateMixtures') iosr.bss.generateMixtures>,
|
|
c@0
|
63 % <matlab:doc('iosr.bss.resynthesise') iosr.bss.resynthesise>.
|
|
c@0
|
64 %
|
|
c@0
|
65 % <html>
|
|
c@0
|
66 % <hr>
|
|
c@0
|
67 % <p>Copyright © 2017 <a href="http://www.surrey.ac.uk">University of Surrey</a><br>
|
|
c@0
|
68 % <a href="http://iosr.uk">Institute of Sound Recording, University of Surrey, UK</a></p>
|
|
c@0
|
69 % </html>
|
