How to generate and add Gaussian white noise in speech signal processing

Digital technology has a far-reaching impact on today's almost digital world. Digital audio technology is called widespread use. Digital technologies such as CD and VCD have already entered the digital broadcasting of thousands of households, and gradually developed some products and applications closely related to global consumers. The topics to be introduced in this paper are: the emergence and development of digital audio compression technology, and the present and early application of audio compression technology. Audio compression technology refers to the technology of using digital signal processing technology to lose the information of the original digital audio signal stream (pulse code modulation) or ignore the loss condition (compression). Its rate compression coding must have corresponding inverse transformation. Decoding or decoding audio signals can introduce the advantages of quantization noise and fixed distortion digital signals. Obviously, its corresponding disadvantages are the increase of storage capacity demand and transmission channel capacity demand. Like CDs. The sampling rate is 44. The quantization accuracy of one kilohertz is one log bit, and a clock stereo audio signal needs to occupy about 10m bytes of storage capacity. It is said that the capacity of a CD turntable is about a high bandwidth, and the problem in the field of digital video is even more serious. It is found that all bits are necessary. It is found that PCM code stream is directly used for storage and transmission, and the sound is compressed at least four to one, that is, 25% of the digital quantity is retained, and the compression ratio in the video field is even hundreds of times. Pay close attention to the research and application of audio compression technology. The long-standing research and development of A-law and U-law coding, simple quasi-instantaneous companding, ISDN sound transmission and the application of voice signals have been well developed and widely used, such as the application of adaptive differential PCM(ADPCM), linear predictive coding (LPC) and other technologies in the broadcasting field, and the audio compression technology II has been adopted in NICAM (near-instantaneous companding audio multiplexing) and other systems. Audio compression is the main category and typical representative. Generally speaking, audio compression technology has lossy compression and lossy compression. According to the compression situation, it can be divided into different compression technologies such as domain compression, transform compression, band compression and mixed compression. Its complexity (including inter-frame complexity and spatial complexity), audio quality, calculation efficiency (compression ratio) and codec delay are the same as those in various compression technology applications. (1) domain compression (or waveform coding) refers to the direct processing of audio PCM code stream values, and the code stream is compressed by means of mute detection, nonlinear quantization and difference. * * * Compression ratio with low computational complexity and sound quality (CD sound quality >: 400 kbps) Coding and decoding delay (phase-to-phase technology) Compression technology is generally used for speech compression low bit rate applications (source signal bandwidth). Domain compression technologies mainly include G. Paint One One, ADPCM, LPC, CELP and other technologies developed from compandor NICAM, and ADPCM(SB-ADPCM) technologies such as G. Paint Two One, G. Paint Two Two, Apt-X, etc. (2) The theoretical basis of band coding with compression technology. The basic idea of strip coding theory has always been that Crochiere is equal to 1-9. The signals in several frequency bands are solved, and each frequency band adopts the same compression strategy to reduce the code rate according to its same distribution characteristics. In the technical introduction of band compression technology, the quantization order of band value or frequency domain value is determined according to the spectrum analysis of communication signal of sound signal knowledge model (sound theory model), and its parameters are selected as perceptual. ) compression coding two compression phase domain compression technologies are more complicated, and the same coding efficiency and sound quality amplitude increase, and the coding delay also increases accordingly. Generally speaking, the complexity of band coding is slightly lower than that of transform coding, and the delay is shorter. Because the band compression technology mainly uses the sound masking model signal to compress, it introduces quantization noise, decodes some noise according to the auditory masking curve, and masks the human ear with the sound signal. Because the frequency band analysis uses the noise in each frequency band to limit the frequency band signal, it affects the quantization order of each frequency band. The compression efficiency is the same as the status bit matching technology. The original bit rate conditional technology realizes completely transparent sound quality (EBU sound quality standard). At present, band compression technology is widely used in the storage and production of digital sound programs. Typical representatives of digital broadcasting are the famous MPEG- 1 layer I and Layer II (Musicam) and their applications in Philips DCC PASC (Precision Adaptive Subband Coding). (3) Transform compression technology and frequency band compression technology are in the same technical segment, which quantize the transform domain parameters obtained by linear transformation, and decompose the transmission signal into several frequency bands by using transform DFT and DCT (Discrete Cosine Transform). MDCT et al. allocated the state bits reasonably according to the parameters of the short power spectrum transform domain of the signal, which significantly improved the audio quality. At the corresponding cost, the computational complexity increases, and there are some defects in transform domain compression, such as block boundary influence, pre-response, and serious degradation of low bit rate sound quality. With the development of technology, some defects are gradually eliminated and new compression coding technology is adopted. Some representative transform compression coding technologies such as DolbyAC- II and AT & amp； Taspec (audio spectrum perceptual coding), PAC (perceptual audio encoder), etc. Standardization of audio compression technology MPEG- 1 Because digital audio compression technology has a wide range of applications and good market prospects, some famous research institutions and companies spare no effort to produce patented products. Standardization of audio compression technology is very important. The standardization of CCITT (now ITU-T) speech signal compression has been completed. Standards such as G. Paint one, G. Paint two, G. Paint two bars, and the standardization of audio compression are gradually recognized by the industry. MPEG- 1 Audio (ISO/IEC111Paint Two-Three) MPEG-1Audio compression specifies three modes, namely, the first layer, the second layer (that is, MUSICAM is called MP II) and the third layer (called Due to the development of standard allowable compression technology, the realization (complexity) of three modes is widely used. VCD uses audio compression case MPEG. MUSICAM is widely used in the production, exchange, storage and transmission of digital programs, such as digital studios, DAB and DVB, because of its excellent sound quality and appropriate complexity. MP 3 combines the advantages of MUSICAMASPEC, and puts forward the technical conditions of mixed compression technology. The complexity of MP 3 is obviously better than that of high coding. Because of the low bit rate and high-level sound quality of MP 3, soft decompression and China network broadcasting pets talk about the formulation and even influence of MPEG- 1 audio standard. Let's start with the MPEG- 2 MPEG- 4 audio standard system.