الفهرس يوجد فقط 14 صفحة متاحة للعرض العام
 |  | from | 111 |  |  |
| | | from | 111 | | |
المستخلص
Digital coding of audio signal is important due to the bandwidth and storage limitations inherent in networks and computers. Audio coding schemes can be divided into lossless and lossy audio coding schemes. Lossless coding techniques involve no loss of information. On the other hand lossy coding techniques involve acceptable loss of information; the information that has been lost may be inaudible such as in perceptual audio coders.
In this thesis, we study the perceptual audio coder that is consisted of three building blocks, namely, a filter bank, a psychoacoustic model 1 and quantization and coding. The filter bank stage transforms the signal into a ftequency domain representation. The psychoacoustic model I applies a high ftequency resolution transform and then applies rules ftom psychoacoustics to calculate the ftequency domain masking threshold. The output ftom both the filter bank and psychoacoustic stage then goes to the quantization and coding stage where the actual bit rate reduction occurs. The coding and quantization stage decides how bits are al10cated among the filter bank coefficients and a quantizer used to quantize the filter bank coefficients. And lossless coding step is applied at this stage to remove statistical redundancy.
Lossy audio coding using overcomplete representation such as best orthogonal basis (BOB), matching pursuit (MP), basis pursuit (BP) and method of ftames (MOF) where the length of output coefficients greater than the length of the input and the compression is made by canceling number of coefficients that have low amplitude.
Lossless coding techniques using integer transforms such as integer Wavelet transform (IntW1), integer Walsh Hadamard transforms (IntWHT) and integer Discrete Cosine transform (IntDCT). In this method the audio signal under consideration, which is assumed to be integer-valued is first decorrelated using the appropriate integer transform. The resulting integer coefficients are then entropy coded by Arithmetic and Huffinan coding to produce the output stream. The performed simulation provides insight on the performance of the different integer transforms in the lossless audio coding context. And compare the results with differential encoding method