Sub band project

• 1. WELCOME

2. GUIDE:MIS.AMBIKA SEKHARGROUP MEMBERS:ATHIRA.PSIRAJ SIDHIKSHAHANA.P.N 3. PROBLEMS speech coding systems is to transmit speech with thehighest possible quality using the least possiblechannel capacity. To save bandwidth in telecoms applications and toreduce memory storage requirements. Maintain certain levels of complexity to reduce theprocessing delay and cost of implementation. 4. PRESENTATION OUTLINE Section I Introduction to speech Sub-band coding (SBC) Filter Banks Section II Sub band coder implimentation QMF design Simulation and result Section III Conclusion Applications 5. Introduction to SpeechWhat is the Speech?o Speech is the primary method of humancommunication.oTo transmit/store a speech waveform using asfew bits as possible while retaining high quality 6. Speech Process Production Propagation: Perception:The incoming sounds are deciphered by the listener into a received message, thereby completing the chain of events that culminated in the transfer of information from the speaker to the listener 7. SUB BAND CODING Divides the speech signal into many smaller sub-bandsand encodes each sub-band separately according tosome perceptual significance. Speech is typically divided into 4 or 8 sub-bands by abank of filters. Can be used for coding speech at bit rates in the range9.6 kbps to 32 kbps. 8. A compression approach where digital filters are used to separate the source output into different bands of frequencies. Each part then can be encoded separately. 9. FILTERS A system that isolates a constituent part corresponding to certain frequency is called a filter. If it isolates the low frequency components, it is called alow- pass filter. Similarly, we have high-pass or band pass filters. In general, a filter can be called a subband filter if it isolates a number of bands 10. FILTER BANKS Filter banks are essentially a cascade of stages, where eachstage consists of a low-pass filter and a high-pass filter The source output is passed through a bank of filters. This filter bank covers the range of frequencies that makeup the source output. The passband of each filter specifies each set offrequencies that can pass through. 11. FILTER BANKS 12. SUB BAND CODER IMPLIMENTATION 13. MATLAB CODE IMPLIMENTING THE SUBBAND CODERFunction y=subband(x,h0,bits) subband decomposition y=subband(x, h0, [bits]) x=input signal vectorh0=basic QMF filter bits= a vector of 2 entries giving the number of bits y=output signal vector 14. SUB BAND CODING ALGORITHM 15. 1.ANALYSISBLOCK DIAGRAM OF A SUB BAND SPEECH ENCODER WITHTHREE FREQUENCY SUBDIVISION 16. The speech signal is to be sampled at a rate fs samplesper second. The first frequency subdivision is splits the signalspectrum into two equal width segments,low passsignal and a high pass signal The second frequency subdivision split the firstlowpass signal into two equal bands ,a lowpasssignal,,,and a highpass signal 17. Finally, the third frequency subdivision splits thelowpass signal from the second stage into two equalbandwidth signals . Thus the signal is subdivided into four frequencybands,covering three octaves. 18. BLOCK DIAGRAM OF SUB BAND SPEECH DECODER WITH THREEFREQUENCY SUBDIVISION 19. The decoding process for the sub band encodedspeech signal is basically the reverse of the encodingprocess.The signal in adjacent lowpass and high passfrequency bands are interpolated, filterd,andcombined 20. Quadrature Mirror Filter (QMF)A quadrature mirror filter is a filter most commonly used toimplement a filter bank that splits an input signal into twobands. The resulting high-pass and low-pass signals are oftenreduced by a factor of 2, giving a critically sampled two-channel representation of the original signal. 21. DECIMATION Downsampling (or "subsampling") is the process of redusing the sampling rate of asignal. This is usually done to reduce the data rate or the size of the data. 22. INTERPOLATOR Upsampling is the process of increesing the sampling rateof a signal. The upsampling factor (commonly denoted by L) isusually an integer or a rational fraction greater than unity. 23. 2.Quantization and CodingSelection of the compression schemeAllocation of bits between the subbandsAllocate the available bits among the subbands according to measure of the information content in each subband. 24. Bit AllocationMinimizing the distortion i.e. minimizing the reconstruction error drives the bit allocation procedure.Bit allocation procedure can have a significant impact on the quality of the final reconstruction 25. 3.Synthesis Quantized and Coded coefficients are used to reconstruct arepresentation of the original signal at the decoder. Encoded samples from each subband decodedupsampled bank of reconstruction filters outputscombined Final reconstructed output 26. SIMULATION AND RESULTS 27. CONCLUSION Subband coding is another approach to decompose thesource output into components based on frequency. A structure of two channel QMF with lowpassfilter,highpass filter,decimators and interpolators hasbeen proposed to perform subband coding of speechsignal in the digital domain. 28. The general subband encoding procedure can be summarizedas follows: Select a set of filters for decomposing the source. Using the filters, obtain the subband signals. Decimate the output of the filters. Encode the decimated output. The decoding procedure is the inverse of the encodingprocedure 29. APPLICATIONS Speech Coding ITU-T G.722 Encode high quality speech at 64/56/48 kbps Audio Coding MPEG audio Image Compression 30. REFERENCES YUE Dongjian The Study of Speech Coding TechnologyBased on Code Excited Linear Predictive CodingPh.D.thesis, Tongji University, 2000. B. Carnero and A. Drygajlo. Perceptual speech codingand enhancement using frame synchronized fast waveletpacket transform algorithms. IEEE Trans. SignalProcessing Vol.47 No.6 ,June 1999. P. Philippe, F. Moreau de Saint-Martin and M. Lever.Wavelet packet filterbanks for low time delay audiocoding. IEEE Trans. Speech and Audio Processing. 1999. 31. John G. Proakis and Dimitris G. Manolakis, DigitalSignal Processing: Principles,Algorithms andApplications, Third Edition. Roberts R. A. and Mullis C. T. Digital Signal Processing.Addison-Wesley, Reading. Mass, 2006. [3]. Oppenheim A. V. and Schafer R. W. Discrete-TimeSignal Processing. Prentice Hall. Englewood Cliffs, New Jersey, 2007. 32. QUESTIONS???