rework algorithm selection

[ThomasFeher]

Target:

• easier integration of new algorithms
• calculation of multiple algorithms during one run (even parallel in later version)
• resource saving by knowing at start what pre- and post-processing steps will be needed (i.e. frequency domain transformation)

Implementation:

• each algorithm needs a properties structure, that can be obtained with a special function getProp<Algorithm>
  • properties are:
    • function handle
    • needs frequency domain signal
    • needs audio input
    • result position in result structure
    • number of audio input signals
    • can run in evaluation mode (non-adaptive)
• the do<Algorithm> key for each algorithm is replaced with one algorithms cell-array containing the names of all algorithms to be run
• based on the list, the framework can obtain the algorithm property structure, and based on that, it can call the algorithm and do all necessary preparations (i.e. calculate frequency domain signals)
• transformation back to time domain can search through the list and the associated properties to find the algorithms that need transformation and the position of their output audio data
• additionally a evalAlgorithms could contain a list of algorithm that should be evaluated (means run non-adaptiv on the separate input signals)
  • needs the algorithm to be able to do that
  • needs convolution used to generate input signals (to get access to the separate signals)
  • needs the algorithms to be run in normal mode beforehand, therefore be in the algorithms list

Branch: algorithm_selection

[ThomasFeher]

keep in mind that it would be nice to be able to also process the same algorithm with different configurations at once