Design Of Low Delay Decimation For HD Audio

Sigma-Delta ADC can obtain high-definition audio digital audio signals due to its high precision,but its sacrifice of speed for high precision greatly increases recording delay.When recording audio,if the difference between the sound that the singer transmits to the ear through the skull and monitor headphones is greater than 1 millisecond,the singer will feel uncomfortable.Therefore,the delay caused by the existing Sigma-Delta ADC often becomes the bottleneck of high-quality audio acquisition and production.The most complicated part of Sigma-Delta ADC calculation is decimation filter,which mainly affects the performance of Sigma-Delta ADC such as energy consumption and delay.How to reduce the delay of decimation filter and improve the audio quality has become a problem that must be solved in this field.This thesis mainly studies the decimation filter of low delay and high definition audio recording equipment.The existing work reduces the energy consumption of decimation filter by using multi-stage FIR filter.Some improvements in the theory of multi-stage filter hierarchical design are also made.The design idea is to shorten the width of the transition band of the multistage filter,i.e.to remove the normalized transition band width of other stage filters in the last stage,so as to reduce the filter coefficient N.The reduction of N also means that the evaluation performance of the filter reduces,such as group delay,energy consumption and chip footprint.The fast transition band roll-off in IIR is used to reduce the overshoot caused by the narrow bandwidth design of hybrid multistage filters,so that the half-band filter has a wider transition band.Compared with the classical multistage filter,the design method in this thesis has a wider transition band design,so the method reduces the filter group delay while reducing the energy consumption of the filter on the premise of no loss in other performances.In this thesis,experiments are carried out to obtain the results under MATLAB simulation on behalf of universal multistage filter(Equiripple),special filter(half band filter)and mixed half band multistage filter.The three multistage filters are selected according to different grading multiples of the filters.The experimental results show that the MWTB algorithm in this thesis can reduce the group delay and energy consumption of the filter by 0 ~ 3% regardless of the grading multiple.IIR design is added to the design of hybrid multi-stage filter,which reduces the height of the blocking and overshoot region by about 5dB,further reducing the limitation brought by the special frequency domain curve of half-band filter.The half-band filter improves the performance of the filter as a whole due to its fast computation.