To merge the reconstructed and existing signals together, a number of issues need to be resolved. First, due to the fact that the reconstructed signal may differ in amplitude from the original, an amplitude correction algorithm needs to be applied to the reconstructed signal. Also, synchronization between the actual wave and the newly reconstructed wave needs to occur. Lastly, normalizing the harmonics against the background noise is necessary to allow the new harmonics to be properly heard.
Except for synchronization, there are two primary parameters that substantially affect how the merge is performed. Those parameters are the “rate of averaging” and “degree of smoothness”. The “rate of averaging” applies to rate of change the merged sound can have based on the previous sounds merged within the engine. Its importance applies to the quality of sound being processed. Highly damaged sound waves perform poorly under high rates of change whereas less corrupt sound waves perform exceedingly well with the same value. The “degree of smoothness” adjusts how the harmonic energies of the genes themselves are allowed to behave. High values of “smoothness” normalize the behavior of the sound gene harmonics. Low values of “smoothness” allow abrupt changes in high frequency harmonics making the sound appear more “bright”. Again, the sound quality plays an important part on how this value is determined. Only exact combinations of all parameters allow the proper balance of improvement and continuance. These algorithms and their parameters, along with synchronization, perform the merging of the signals giving the original signal the additional fidelity.
<<previous [ 1 ] [ 2 ] [ 3 ] [ 4 ] [ 5 ] [ 6 ] 7 [ 8 ] [ 9 ] [ 10 ] [ 11 ] [ 12 ] next >>
Download full copy of Sound Genome Theory white paper in pdf format. 