Compression and Decompression for Multimedia Systems.
James Rytter CCM 3101 2116104
Compression and Decompression for Multimedia Systems
Methods of compression techniques in existing voice mail systems
There are many different ways to compress a file containing voice data. These methods include LPC, LPC-10, CELP, MP3, MP4. All are lossy methods, this means that some of the original data has been removed in a way that the end result will not sound much different from the original recorded wave form. This method works in a similar way to that of JPEG picture compression. Irrelevant data is removed, this could be silences or removing higher or lower frequency ranges then the human ear can distinguish, doing so makes the resulting file size a lot smaller.
LPC ‘Linear Predictive Coding’ gives the best compression of all available algorithms, as it reduces the data rate by more than a factor of 12. This method gives the greatest degree of compression of any of the current available techniques. The down side of this type of compression is that it is very demanding on the maths co processor due to the amount of floating point computations that it must perform to convert the sound file. If the machine that the converting does not have a math co processor it would take a very long time to complete a conversion, or in most cases be impossible to complete in real time. LPC is very sensitive to high frequencies and clipping caused by a microphone with an input level set to high, this is even more apparent when a person with a high frequency voice uses the compression as LPC removes a lot of the high frequency data, making their message (in the case of voice mail) unintelligible.