EURASIP Journal on Audio, Speech, and Music Processing

Table 11 Statistics of bits per syllable of the proposed approach and the baselines with various logF0 codebooks for the SD speech coding case

From: A parametric prosody coding approach for Mandarin speech using a hierarchical prosodic model

	Bits per syllable
	Average			Standard deviation		Maximum/Minimum
	NC	M0	M1	M0	M1	M0	M1
(a) Inside
SD-HPM	27	23.23	20.65	0.30	0.38	24.23/22.57	22.09/19.59
SD-BSL-24	25	20.95	20.12	0.20	0.32	21.84/20.44	21.21/19.16
SD-BSL-32	25	21.29	20.45	0.20	0.32	22.14/20.73	21.52/19.35
SD-BSL-64	26	22.31	21.20	0.20	0.31	23.20/21.81	22.14/20.24
SD-BSL-128	27	23.09	21.83	0.20	0.32	23.96/22.54	22.71/20.78
SD-BSL-256	28	24.10	22.12	0.20	0.32	24.95/23.55	22.93/21.14
(b) outside
SD-HPM	27	23.04	21.42	0.29	0.39	23.72/22.45	22.31/20.48
SD-BSL-24	25	21.31	21.01	0.19	0.34	21.96/21.02	21.62/20.19
SD-BSL-32	25	21.59	21.38	0.20	0.37	22.25/21.26	22.05/20.35
SD-BSL-64	26	22.65	22.72	0.21	0.56	23.37/22.34	23.82/21.48
SD-BSL-128	27	23.59	27.70	0.19	1.61	24.29/23.28	31.69/24.63
SD-BSL-256	28	24.64	54.37	0.21	6.60	25.37/24.31	67.40/41.36

Back to article page