Audio Subsystem Design And Audio Signal Processing Algorithm Research For Digital Television

Digital Audio is a core component in DTV,because audio is central to perceived quality of DTV product.With the development of integrated circuits.DTV can be integrated in the system-on-chip.The chanllenge of DTV SoC design is to integrated more features with lower cost.The audio subsystem design is of no exception.This thesis makes research on the field of Digital Audio Technique in DTV decoding SoC.Based on the design of demultiplexer and audio subsystem,a RISC-based audio processor is adopted for multi-standard decoding and processing in a new gerneration DTV SoC.On its basis,serveal efficient audio processing algorithms are proposed in MDCT domain,which is the dominant transform method of high quality audio coding in DTV.The achievements are as follows:1,Proposed an efficient Audio-video synchronization method in the design of demultiplexer and audio subsystem in the DTV SoC.As an important part in SoC chip of DVB-S demodulator and MPEG-2 decoder,all the design was verified by ASIC and mass production.2,Proposed an accurate low complexity algorithm for frequency estimation in MDCT domain,The algorithm extracts integer part of frequency by estimating combinations of local maximum and neighbor of MDCT coefficients.And the fractional part of the frequency achieves accurate results by complementing method. The algorithm is superior to previous method in Sinusoid modeling of MDCT domain.3,Proposed an efficient audio packet loss concealment method in MDCT domain. Based on Sinusoid and noise model,MDCT coefficients are divided into two groups, noise and sinusoid component.The sign of MDCT coefficients of sinusoidal component in the lost packets are reconstructed by frequency estimation.The method can improve the subjective listening quality with quite low computational cost,so can be used in DTV receiver with limited computation capacity.4,Proposed an efficient filtering approach in MDCT domain base on the spectrum character of HRTF.On its basis,realized a virtual surround processing method in MDCT domain.Compared with previous frequency and MDCT domain methods,the proposed virtual surround processing achieves low complexity with the same quality,both in objective and subjective test.