Compression of audio signals.
Audio compression is widely used in professional and consumer digital audio products, such as compact disc (CD), digital audio type (DAT), mini disc (MD), digital compact cassette (digital compact cassette – DCC), versatile disc digital (DVD), digital audio broadcasting (DAB) and MP3 audio products from M <Picture Experts Group – (MPEG).
While voice compression in telephony, in particular cellular telephony, necessary to save bandwidth and save time and battery, has led to the development of many voice compression standards, personal algorithms are applicable to voice signals and consumer of a wider frequency band. Voice and audio compression schemes can be conveniently classified according to applications, reflecting some measure of acceptable quality.
Adaptive Differential PCM (ADICM).
Using the past data to measure (i.e., quantify) the new ones, we went from conventional pulse code modulation (PCM) to differential (differential PCM – DPCM). In DPCM, a prediction of the next sampled value is generated based on the previous values. The quantizers are called instantaneous quantizers or non-memory quantizers because the digital transformations are based on a single (current) input sample. These properties were unequal source levels and dependent sample values. The correlation characteristics of a source can be represented in the time domain by sampling its autocorrelation function and in the frequency domain by its power spectrum. If we study the power spectrum Gx (f) of a short-term speech signal, as shown in Figure 9.2, then we see that the spectrum has a global maximum in the vicinity of 300 to 800 Hz and decays at a rate 6 to 12 dB / octave. This operation is performed on the legend and the comparison contour, the upper contour of the encoder is shown in Figure 13.2. The encoder adjusts its predictions by adding the predicted value and the prediction error.
This model, which uses a 12-lead speech synthesizer, has found application in children’s conversation games.
Compression algorithm
MPEG
The International Organization for Standardization (ISO) and the Motion Picture Experts Group (MPEG) have developed the audio compression standard for synchronized video signals known as MPEG. This scheme combines the properties of MUS1CAM (Masking Pattern Adaptive Universal Subband Integrated Coding and Multiplexing) and ASPEC (Adaptive Spectral Perceptual Entropy Coding). The scheme uses three levels (codes) of increasing complexity and improvement of subjective performance. The input sampling frequencies are 32, 44.1 and 48 kHz, and the bits are output at 32 to 192 kbps (monaural) or 64 to 384 kbps (stereo). The standard supports single channel mode, stereo mode, dual channel mode (for bilingual audio programs), and optional collaborative stereo mode. In the latter mode, the two encoders for the left and right channels can support each other using common statistics to reduce the bit rate of the audio signal even more than is possible with mono transmission.
Level III of the MPEG / ISO (MP3) standard achieves a higher frequency resolution that is very close to critical human resolution.
