Application of 1-D discrete wavelet transform based compressed sensing matrices for speech compression

SpringerPlus

Table 1 Performance analysis of the proposed db1 (Haar) wavelet based sensing matrix

Length of signal (N)	Number of measurements (m)	Compression ratio (CR = m/N)	Sparsity level = (k/N) × 100 (%)	No. of non-zeros (k)	No. of iterations required	Signal reconstruction time (s)	RMSE	Relative error	SNR (db)	Construction time for sensing matrix (s)
2048	205	0.1	50	1024	16	1.281382	0.0585	0.9774	0.1985	2.584058
2048	410	0.2	50	1024	16	2.331942	0.0521	0.8711	1.1984	2.523152
2048	512	0.25	50	1024	16	3.752369	0.0521	0.8707	1.2024	2.514315
2048	614	0.3	50	1024	16	4.242975	0.0488	0.8153	1.7740	2.570737
2048	849	0.4	50	1024	14	5.853003	0.0479	0.7997	1.9418	2.581421
2048	1024	0.5	50	1024	14	9.486588	0.0478	0.7993	1.9454	2.581856
2048	1229	0.6	50	1024	14	11.080416	0.0471	0.7877	2.0723	2.593704
2048	1434	0.7	50	1024	13	14.440796	0.0468	0.7824	2.1310	2.546346
2048	1536	0.75	50	1024	12	17.799127	0.0468	0.7816	2.1406	2.523620
2048	1638	0.8	50	1024	11	16.568279	0.0468	0.7816	2.1406	2.484885
2048	1843	0.9	50	1024	11	20.629636	0.0468	0.7816	2.1407	2.518902
2048	2048	1.0	50	1024	9	22.525032	0.0468	0.7815	2.1409	2.612855