Audio coding
There are many different ways to store digital audio.
The digitized sound is a set of signal amplitude values,
taken at regular intervals.
PULSE-CODE MODULATION
Let us set aside for recording one value of the signal amplitude in memory
computer N bits. Hence, with the help of one N-bit word one can describe
2
𝑁 different meanings. Let the amplitude of the digitized signal fluctuate in
within the range from 0 to 1 of some conventional units. Imagine this range
amplitude changes in the form of 2
Equal intervals. Now, for everyone’s record
separate amplitude value, it must be rounded to the nearest
quantization level (see quantization noise). This process is called
quantization in amplitude (level). Amplitude quantization – process
replacement of real values of the signal amplitude with values approximated to
some precision. Each of 2
𝑁 possible levels is called a level
quantization, and the distance between the two nearest quantization levels
called the quantization step. If the amplitude scale is divided into levels
linearly, then the quantization step is ∆ =
1
2𝑁
… This way of digitizing a signal –
sampling of the signal in time in conjunction with the homogeneous method
quantization – called linear pulse code modulation (PCM)
(Linear Pulse Code Modulation – LPCM) or just PCM (PCM).
To record sound on digital media or transmit via communication channels (see.
SPDIF interface) data in parallel code are encoded when
shift register, clocked by an auxiliary generator. At the exit
shift register, packets of coded pulses are formed in
sequential code.
2
Standard Audio CD (CD-DA) since the early 1980s
years of the 20th century, stores information in PCM format with a frequency
44.1 kHz sampling and 16-bit quantization.
There are varieties of PCM that use uneven
quantization step (Nonuniform PCM). Non-uniform (non-linear) method
quantization provides for dividing the amplitude scale into levels, according to
nonlinear (usually logarithmic) law. This way
quantization is called logarithmic quantization. Using
logarithmic amplitude scale, in the region of weak amplitude it turns out
a greater number of quantization levels than in the region of strong amplitude (in this case,
the total number of quantization levels remains the same as in the case of a uniform
quantization). The relative error is const.
An alternative A / D conversion method is
differential (difference) pulse-code modulation DPCM
(Differential PCM). In the case of DPCM, it is not the amplitude itself that is quantized,
and the difference between the current and the previous measured. In complete analogy with
PCM, difference PCM can be combined using both uniform and
and uneven quantization. For audio data, this type of modulation
reduces the required number of bits per count by about 25%.
3
Difference coding has many different options. For example, adaptive
DPCM (ADPCM, ADPCM) – a kind of DPCM with variable pitch
quantization. Changing the pitch allows you to reduce bandwidth requirements
transmission at a given signal-to-noise ratio.
A block of digitized audio information can be written to a file without
changes, that is, a sequence of amplitude values describing
audio waveform, i.e. uncompressed. For this it is usually used
.WAV format. Waveform Audio File Format (WAVE, WAV, from waveform) –
a container file format for storing a recording of a digitized audio stream,
RIFF subspecies. This container is typically used to store uncompressed
sound in pulse-code modulation. However, the container does not impose any
restrictions on the used encoding algorithm
