Sub-band Coding (SBC)

The basic idea of the

Subband encoding (Subband CODING,SBC) is to use a set of Bandpass filters (Band-pass FILTER,BPF) to divide the frequency band of the input audio signal into several successive bands, each of which is called a subband. A separate coding scheme is used to encode the audio signals in each sub-band. When transmitting on a channel, the code of each child band is composited together. At the receiving end of decoding, the code of each sub-band is decoded separately and then combined to restore the original audio signal. The sub-coded block diagram is shown in Figure 3-17, the code/decoder in the figure can be used ADPCM,APCM,PCM and so on.
Figure 3-17 Sub-coded block diagram

The

has two benefits of encoding each sub-band separately. First, each sub-band signal is individually adaptive control, the size of the quantization order (quantization step) can be adjusted according to the energy level of each sub-band. A sub-band with a higher energy level is quantified with a large quantization order to reduce the total quantization noise. Second, according to each sub-band signal in the sense of importance, each sub-band assigned a different number of digits, used to represent each sample value. For example, in the low frequency sub-band, in order to protect the tone and resonance peak structure, it is required to use a smaller quantization order, more quantitative series, that is, allocating more digits to represent the sample value. In the voice, the friction sound and the similar noise sound, usually appears in the high frequency sub-band, assigns it the small number of digits. The segmentation of the
Audio band can be divided by the pattern of the tree structure. First, the entire audio signal bandwidth is divided into two sub-bands of equal bandwidth: high-frequency sub-band and low-frequency sub-band. The two sub-bands are then divided in the same way, forming 4 sub-bands. This process can be repeated as needed to produce a 2K sub-band, K is the number of splits. In this way, a sub-band with equal bandwidth can be produced, and a sub-band with unequal bandwidth can also be generated. For example, for an audio signal with a bandwidth of 4000 Hz, when k=3, it can be divided into 8 sub-bands of equal bandwidth, each with a bandwidth of at most Hz. You can also generate 5 sub-bands of unequal bandwidths, respectively [0,500], [500,1000], [1000,2000], [2000,3000], and [3000,4000]. After the
divides the audio signal into the neighboring sub-band component, it can reconstruct the original sub-band signal with its sample value by sampling the signal with a sampling frequency of twice times the bandwidth of the sub-band. For example, when the 4000 Hz bandwidth is divided into 4 equal bandwidth sub-bands, the band bandwidth is three Hz and the sampling frequency is available in Hz, and its total sampling rate is still 8000 Hz.
because the filter used in the segmented band is not an ideal filter, the output audio signal synthesized after striping, encoding and decoding will have a aliasing effect. According to the analysis of the data, using the orthogonal mirror image filter (quandrature mirror FILTER,QMF) to divide the frequency band, the aliasing effect can be offset at the final synthesis.
Figure 3-18 shows a simplified block diagram of the sub-band compiler code with the QMF split band. The figure shows the use of QMF to divide the full-bandwidth audio signal into two equal bandwidth sub-bands. HH (n) and HL (n) represent high-pass filters and low-pass filters, respectively, which form a pair of orthogonal mirror-image filters. These two filters are also called analytical filters. Figure 3-18 (b) is a simplified amplitude-frequency characteristic of the QMF.
(a) QMF split channel block diagram

(b) simplified diagram of amplitude-frequency characteristics of QMF

Figure 3-18 Simplified block diagram with QMF code for the sub-band

The sub-band encoder SBC is gaining more and more attention. In the medium rate coding system, SBC has wide dynamic range, high sound quality and low cost. Codec using sub-band coding technology has been used for voice storage and forwarding (voice store-and-forward) and voice mail, the use of 2 sub-band and ADPCM encoding system has been CCITT as the g.722 standard recommended to the world.