CN1144174C - Method for discriminating acoustic figure with base band components and sounding parameters - Google Patents

Abstract

The present invention relates to a new parameter in voiceprint identification and a new method. An analysis system which mainly comprises an acoustic parameter voice database which reflects human body individual differences is used to execute judgement by using a combination of characteristic acoustic parameters. The present invention is characterized in that fundamental frequency component phase differences of voice syllables are used as a characteristic acoustic parameter, and the characteristic acoustic parameter is combined with another characteristic acoustic parameter during sound production for voice measurement. The method of the present invention is simple and practical and can be used for judiciary voiceprint identification, security, sound locks, money systems and clinical vocal cord obstacle diagnosis.

Description

The vocal print identification method of parameter when adopting fundamental component phase differential and sounding

Affiliated field: the present invention relates to new physiological psychology parameter and assay method thereof in the individual identification of the person.Specifically, be new parameter and new method in a kind of vocal print identification.

Background technology: U.S. Department of Defense assigns Bell Laboratory about after analyzing the research task of supreme command of German army session recording material with definite words person's name since nineteen forty-one, and countries in the world have been studied the speaker identification of more than 20 kind of acoustical parameter over more than 60 year.Particularly the vocal print identification technology that adopted since the age of 60-70 is used in many fields such as various countries' administration of justice, security personnel, but this technology has mistake still the time.

With United States Patent (USP) 6,029,124, " Sequential, Nonparametric Speech Recognition andSpeaker Identification " (Gillick, et al., February 22,2000) to compare, this patent adopts traditional characteristic parameter.Following table has reflected " performance of characteristic parameter combination relatively ".

Table: the performance of characteristic parameter combination relatively

The characteristic parameter misclassification rate, %

Cepstrum 9.43

Difference cepstrum 11.81

Fundamental tone 74.42

Difference fundamental tone 85.88

Cepstrum and difference cepstrum combination 7.93

Cepstrum difference cepstrum 2.89

The group of fundamental tone difference fundamental tone and

Be not difficult to find out the superiority of characteristic parameter combination from last table; Still have an appointment about 10% misclassification rate of every parameter even above-mentioned five parameters are carried out multiple combination, still has 2.89% misclassification rate but simultaneously as seen.This shows, need seek new parameter the task of speaker's discriminating and affirmation.

Because the classical acoustic notional result thinks that tone color is decided by sound different frequency composition and intensity thereof, and irrelevant with phase propetry.So, studied 20 surplus kind of acoustical parameter all do not relate to phase propetry.But find in the research of phonetic synthesis that when 2-3 pure tone was mixed into complex tone, phase propetry and compound tone color were closely related therebetween.By analyzing the phase differential between words person's voice composition, comprise the phase differential between adjacent composition in fundamental frequency, first resonance peak, second resonance peak, found that phase place random variation between the adjacent composition of two resonance peaks and irregular following, but adjacent composition phase change is that periodic function changes in the Base Band.The driving source characteristic of this explanation Base Band reflection vibration is relevant with the characteristics of individual vocal cords physiological function and vocal cords structure.Therefore this invention works out the assay method that adopts the phase differential between fundamental component, and as measuring everyone vocal cords structure physiological parameter with function, i.e. voice driving source parameter is so that as the new tool of confirming or differentiate person individuality.Since 1967, (Voice On-set Time is called for short VOT) characteristic parameter also is used to the auxiliary parameter of introduction on linguistics research and aphasis diagnosis during sounding simultaneously.Because it can reflect the habituation of channel parameters and individual sounding, promptly therefore channel information also can be used for the acoustical parameter as individual difference.And VOT is the necessary condition of fundamental component measuring difference of phases in the present invention, and itself is again the new parameter of the Physiological Psychology of personal individual difference.Therefore we use new characteristic parameter, and the new parameter as vocal print identification when fundamental component phase differential and sounding has also been invented new vocal print identification method thus.

Goal of the invention: the invention provides a kind of more simple and easy to do voice and measure new method, can be used for judicial vocal print identification, security personnel, sound lock and financial sector and clinical middle vocal cords obstacle diagnosis.

The technical scheme of invention: the present invention adopts fundamental component phase differential and two parameters of VOT of speech syllable, sets up automatic analysis system by the sound bank of corresponding parameter.The vocal print identification analytic system partly is made up of hardware and software, and wherein hardware components comprises microphone, sound card and microprocessor, and software section comprises that cutting sound, sound spectrum, frequency spectrum, language spectrum, phase spectral analysis and result judges software.The phonetic material of intending analyzing can be recorded at the scene, and also the phonetic material that other approach can be obtained changes native system over to, cuts the syllable that is applicable to analysis through cutting sound software from phonetic material to be analyzed, and analysis-by-synthesis is judged by the test data that a plurality of syllables draw.Detailed process is at first to the phonetic material analysis of spectrum of speaking, and thus to speaker's (BA, DA, GA, KA, PA, TA) six basic syllables are analyzed, and measure the VOT characteristic parameter of each basic syllable, carry out the analysis of fundamental component phase differential according to VOT characteristic parameter result, calculate then its individual difference and with database in pattern match, reach at last that the person is individual to be assert.

Beneficial effect: outstanding advantage of the present invention is: on the parameter basis of tradition words person identification, adopt two brand-new parameters to reflect vocal cords respectively, improve discrimination as the individual difference of the physiological function characteristic of driving source and the individual habitual difference of sound channel; This method is simple and easy to do, not only applicable to the phonetic material of specific recording but also can give the processing post analysis through native system with having different phonetic; And a plurality of syllables of analysis-by-synthesis (being generally 6 syllables), can comparatively fast provide test result.

SPSS is social science statistical analysis software bag (Statistical Package for the SocialScience), and this is a statistical analysis software bag that is applicable to natural science, each field of social science, is widely current in countries in the world.ANOVA is a kind of simple variance analysis (Analysis of Variance), can finish statistical study by the Statistics menu call ANOVA in the SPSS master menu in the utilization.

With the ANOVA statistical method in the SPSS software package, the hexasyllable fundamental frequency twenty percent of adding up 102 people respectively divides the individual difference of phase differential to obtain, and the main effect of six syllable individual differences is all remarkable in the .000 level, illustrates that this parameter is very high to the discernment of individual difference.

Table 1,2 shows the VOT average difference of Ba sound and Pa sound, shows that VOT is also very strong as the resolving ability of characteristic parameter.

Table 3 is that the fundamental frequency twenty percent of six syllables of 102 examples divides the phase differential data.

The analysis of six syllable individual differences of 30 people main effect obtains, and the main effect of Ba sound individual difference is remarkable in the .029 level, and other five sounds are remarkable in the .000 level.Six sound individual differences of this 30 people analysis in table 4.Account for 67.0% what the .05 level had a significant difference; What wherein the Ba syllable had a significant difference accounts for 13.8%, what the Da syllable had a significant difference accounts for 15.0%, and what the Ga syllable had a significant difference accounts for 21.8%, and what the Ka syllable had a significant difference accounts for 13.9%, what the Pa syllable had a significant difference accounts for 16.0%, and what the Ta syllable had a significant difference accounts for 19.5%.The discernment that Ga syllable and Ta syllable are described is higher.

The VOT average difference (ms) of table 1,10 tested ba sounds of women

	1	2	3	4	5	6	7	8	9	10
											1	-	1.9 *	-8.1 *	1.1	1.3 *	1.6 *	1.7 *	-0.2	-2.3 *	-4.6 *
2	*	-	-10 *	-0.8	-0.6	-0.3	-0.2	-2.1 *	-4.2 *	-6.5 *
											3	*	*	-	9.2 *	9.4 *	9.7 *	9.8 *	7.9 *	5.8 *	3.5 *
4			*	-	0.2	0.5	0.6	-1.3 *	-5.4 *	-1.7 *
											5	*		*		-	0.3 *	0.4	-1.5 *	-5.6 *	-1.9 *
6	*		*			-	0.1	-1.8 *	-3.9 *	-1.2 *
											7	*		*				-	-1.9 *	-4.0 *	-6.3 *
8		*	*	*	*	*	*	-	-2.1 *	-4.4 *
											9	*	*	*	*	*	*	*	*	-	2.3 *
10	*	*	*	*	*	*	*	*	*	-

Annotate: experiment pronunciation number of times is N=10, ^*It is remarkable to be illustrated in 0.05 level.

The VOT average difference (ms) of table 2,10 tested pa sounds of women

	1	2	3	4	5	6	7	8	9	10
											1	-	14.3 *	-57 *	-42.9 *	-16 *	-49.9 *	-48.3 *	-51.8 *	-47.8 *	-101 *
2	*	-	71.3 *	-57.2 *	-30.2 *	-64.2 *	-62.6 *	-66.1 *	-62.1 *	-115 *
											3	*	*	-	14.1 *	41 *	7.1	8.7	5.2	9.2	-43.8 *
4	*	*	*	-	26.9 *	-7.0	-5.4	-8.9	-4.9	-54.9 *
											5	*	*	*		-	-33.9 *	-32.3 *	-35.8 *	-31.8 *	-84.8 *
6	*	*			*	-	1.6	-1.9	2.1	-50.9 *
											7	*	*			*		-	-3.5	5	-52.5 *
8	*	*			*			-	4	-49 *
											9	*	*			*				-	-53 *
10	*	*	*	*	*	*	*	*	*	-

Annotate: experiment pronunciation number of times is N=10, ^*It is remarkable to be illustrated in 0.05 level

The fundamental frequency twenty percent of table 3, six syllables of 102 examples divides phase differential (π rad)

	ba(mean)	ba(std)	da(mean)	da(std)	ga(mean)	ga(std)	ka(mean)	ka(std)	pa(mean)	pa(std)	ta(mean)	ta(std)
													1	0.381	7.97E-02	0.6067	0.5149	0.3359	4.49E-02	0.1582	6.54E-02	0.8947	0.4494	1.3695	0.262
2	0.3359	4.49E-02	1.124	0.2105	0.4981	0.1311	0.8965	0.2912	1.4429	0.3459	1.6251	0.2783
													3	0.5085	0.2253	1.1951	0.2587	0.3094	0.1309	0.499	0.2084	1.4617	0.2411	1.2207	0.1971
4	0.4809	0.2389	1.3964	0.2358	0.4364	4.71E-02	0.3989	0.2765	1.0307	0.5512	1.289	0.2254
													5	0.3752	0.199	1.2624	0.4448	0.3849	6.31E-02	0.2571	0.27	0.9366	0.544	1.2971	0.2388
6	0.3755	0.1285	0.4683	0.1321	0.577	0.2645	1.1692	0.1538	1.1177	0.2771	1.3517	0.1891
													7	0.5426	0.1185	0.7696	0.4986	0.4732	6.46E-02	1.124	0.2105	0.3074	0.2807	1.2807	0.1235
8	0.4594	0.1433	0.6681	0.4855	0.3391	5.38E-02	1.1951	0.2587	0.8835	0.2224	1.0045	0.2462
													9	0.365	0.1016	0.5133	0.3602	0.3359	0.1236	0.3818	0.2571	1.1794	0.3824	0.3246	0 2073
10	0.53	0.2137	0.4112	0.2946	0.5829	0.2427	1.2008	0.2193	1.3502	0.3682	1.2618	0.3301
													11	0.3262	1.99E-02	0.3214	1.61E-02	0.3359	1.99E-02	1.7361	5.07E-02	1.7507	4.25E-02	1.7506	5.81E-02
12	0.1275	1.93E-02	0.1179	2.02E-02	0.1227	1.90E-02	1.7652	4.24E-02	1.7813	3.96E-02	1.7393	2.02E-02
													13	0.415	1.87E-02	0.4231	1.67E-02	0.4231	1.67E-02	1.8282	1.67E-02	1.8314	1.56E-02	1.8354	3.17E-02
14	0.6638	1.78E-02	0.6557	1.36E-02	0.667	1.87E-02	0.9044	1.08E-02	0.8995	1.09E-02	0.8802	5.18E-02
													15	0.9092	1.87E-02	0.9044	1.52E-02	0.9011	1.67E-02	0.7978	1.36E-02	0.7978	1.13E-02	0.8236	4.31E-02
16	6.78E-02	1.49E-02	7.10E-02	1.36E-02	6.13E-02	1.27E-02	0.8414	1.93E-02	0.8317	2.19E-02	0.8414	1.93E-02
													17	0.3342	2.17E-02	0.3294	2.05E-02	0.3391	2.30E-02	1.0433	8.31E-03	1.04	8.31E-03	1.0905	0.1255
18	0.2051	1.53E-02	0.2018	1.57E-02	0.1986	1.71E-02	1.8637	8.37E-03	1.8621	7.83E-03	1.8459	2.64E-02
													19	0.4845	1.32E-02	0.4893	1.33E-02	0.4796	1.09E-02	1.2484	1.71E-02	1.2387	2.16E-02	1.2338	2.18E-02
20	0.9948	1.90E-02	0.99	1.53E-02	0.9916	1.74E-02	1.3008	1.21E-02	1.2956	8.65E-03	1.2672	7.72E-02
													21	1.0881	1.56E-02	1.0911	1.40E-02	1.0893	1.93E-02	1.111	1.67E-02	1.1029	7.86E-03	1.1478	7.68E-02
22	0.7219	1.33E-02	0.7251	1.61E-02	0.7267	1.52E-02	1.6464	5.37E-02	1.6686	4.50E-02	1.6686	4.50E-02
													23	0.1647	1.27E-02	0.1599	1.41E-02	0.1615	1.32E-02	1.314	2.53E-02	1.3092	2.18E-02	1.3001	2.63E-02
24	0.2939	1.49E-02	0.302	1.71E-02	0.2988	1.75E-02	7.36E-02	1.90E-02	6.72E-02	1.32E-02	8.82E-02	1.33E-02
													25	0.373	1.19E-02	0.3682	1.27E-02	0.3666	1.09E-02	1.94	1.16E-02	1.9335	1.09E-02	1.9338	1.13E-02
26	0.5881	1.87E-02	0.596	1.92E-02	0.596	1.92E-02	2.20E-02	1.83E-02	2.04E-02	1.93E-02	2.04E-02	1.41E-02
													27	0.1857	1.90E-02	0.1776	1.86E-02	0.1744	1.83E-02	0.5234	1.30E-02	0.5191	8.85E-03	0.5311	1.61E-02
28	0.1114	1.19E-02	0.1082	1.09E-02	0.1179	1.33E-02	1.0368	0.1631	1.0206	0.1456	1.1483	0.1483
													29	0.2422	1 32E-02	0.239	1.27E-02	0.2374	1.33E-02	1.51	0.1619	1.4438	0.1002	1.4293	0.1014
30	0.4376	9.17E-03	0.4409	1.09E-02	0.4296	1.13E-02	1.0923	0.1278	1.0665	0.103	1.0371	4.37E-02
													31	0.5362	1.27E-02	0.5313	1.41E-02	0.5394	1.36E-02	1.3001	7.20E-02	1.2759	7.92E-02	1.3243	5.28E-02
32	0.2761	9.17E-03	0.2713	1.02E-02	0.281	1.13E-02	1.7167	0.1946	1.6844	0.18	1.6828	0.206
													33	0.1457	1.37E-02	0.1408	1.40E-02	0.1425	1.54E-02	1.0013	9.45E-02	1.0384	6.09E-02	0.9819	9.59E-02
34	0.4764	1.14E-02	0.4829	1.41E-02	0.4877	1.48E-02	1.334	0.2592	1.2968	0.1745	1.2742	0.2436
													35	0.2245	1.19E-02	0.2212	1.09E-02	0.2325	1.13E-02	1.4035	5.24E-02	1.3841	3.96E-02	1.3841	3.96E-02
36	0.1615	1.08E-02	0.1663	1.09E-02	0.1679	1.13E-02	1.1676	7.54E-02	1.145	7.86E-02	1.1224	7.43E-02
													37	0.2293	4.86E-02	0.2455	3.39E-02	0.2164	5.00E-02	0.5478	1.89E-02	0.5426	1.55E-02	0.5357	1.87E-02
38	0.3456	3.06E-02	0.3488	2.66E-02	0.3327	3.06E-02	1.2306	4.09E-02	1.2048	5.45E-02	1.2435	2.16E-02
													39	0.3569	2.08E-02	0.3472	1.37E-02	0.344	2.29E-02	1.8508	8.45E-02	1.8265	9.58E-02	1.875	6.16E-02
40	0.3908	2.62E-02	0.3843	2.13E-02	0.3747	1.83E-02	1.2887	0.1046	1.2726	0.1027	1.3146	8.72E-02
													41	0.3149	1.37E-02	0.3052	1.61E-02	0.3085	1.61E-02	1.1321	2.69E-02	1.1224	1.74E-02	1.1369	2.43E-02

42	0.3488	2.04E-02	0.3343	1.53E-02	0.3553	1.86E-02	0.1065	6.87E-02	8.55E-02	4.38E-02	0.1286	6.45E-02
													43	0.2784	2.35E-02	0.2716	2.32E-02	0.292	1.68E-02	0.9722	8.94E-02	1.011	8.62E-02	1.011	8.62E-02
44	0.3757	3.14E-02	0.3628	3.04E-02	0.3538	2.95E-02	1.1611	9.58E-02	1.1337	0.1035	1.1111	0.1092
													45	0.237	1.02E-02	0.2412	1.18E-02	0.2342	8.49E-03	1.4244	6.22E-02	1.4567	4.92E-02	1.4002	4.31E-02
46	0.2586	1.45E-02	0.2525	1.05E-02	0.2525	1.51E-02	1.1612	7.20E-02	1.124	5.01E-02	1.137	6.99E-02
													47	0.2487	2.04E-02	0.2487	2.04E-02	0.2584	1.86E-02	0.6993	6.00E-02	0.6783	4.69E-02	0.7154	6.98E-02
48	0.3682	3.56E-02	0.3456	3.06E-02	0.3589	3.53E-02	1.0627	0.18	1.1095	0.1508	1.1085	0.1508
													49	0.3795	1.14E-02	0.3634	2.32E-02	0.3731	2.08E-02	1.4244	6.54E-02	1.4002	4.99E-02	1.3856	4.86E-02
50	0.4651	0.1018	0.4296	8.28E-02	0.4457	9.55E-02	0.9822	0.3096	1.0048	0.3129	1.0643	9.02E-02
													51	0.2842	2.66E-02	0.302	2.29E-02	0.2971	2.66E-02	1.0869	3.65E-02	1.0804	2.99E-02	1.0659	4.69E-02
52	0.3278	3.81E-02	0.2955	3.32E-02	0.3052	3.99E-02	1.8976	6.65E-02	1.8605	6.22E-02	1.8637	5.86E-02
													53	0.3286	1.54E-02	0.3095	2.18E-02	0.3187	2.10E-02	2.69E-02	7.51E-03	2.37E-02	4.84E-03	2.97E-02	6.19E-03
54	0.3267	3.54E-02	0.2957	4.09E-02	0.3037	4.06E-02	1.3846	6.53E-02	1.3636	1.34E-02	1.3636	1.34E-02
													55	0.2424	1.25E-02	0.2325	8.31E-03	0.2383	1.16E-02	1.0239	0.1114	0.9722	9.13E-02	0.9803	0.1047
56	0.2548	1.86E-02	0.2548	1.66E-02	0.2451	1.64E-02	0.9641	0.1055	0.9415	6.81E-02	1.0077	0.1014
													57	0.4377	0.2263	0.3311	8.31E-02	0.4086	0.103	1.258	0.5807	1.258	0.5807	1.4294	0.2946
58	0.3472	6.65E-02	0.3181	7.77E-02	0.4312	9.57E-02	1.345	0.5788	1.345	0.5786	1.4163	0.3245
													59	0.3892	9.88E-02	0.323	6.68E-02	0.394	9.05E-02	1.4758	0.5152	1.4758	0.5152	1.468	0.3212
60	0.4086	0.1276	0.323	8.44E-02	0.4263	0.1083	1.2545	0.5663	1.2545	0.5663	1.397	0.2898
													61	0.4247	0.126	0.3262	6.54E-02	0.4247	0.1069	1.2241	0.5722	1.2241	0.5722	1.3614	0.2577
62	0.4441	0.1315	0.3198	4.56E-02	0.4715	0.1116	1.3048	0.5649	1.3048	0.5649	1.4664	0.2719
													63	0.4344	0.1309	0.3391	5.80E-02	0.4748	0.1197	1.1242	0.6473	1.1242	0.6473	1.4373	0.2554
64	0.428	0.1252	0.3327	7.19E-02	0.4522	8.88E-02	1.4015	0.5113	1.4015	0.5113	1.5342	0.2365
													65	0.4893	0.1292	0.352	7.53E-02	0.4829	0.1094	1.1724	0.581	1.1724	0.581	1.5358	0.296
66	0.4489	0.1669	0.3133	6.69E-02	0.4893	0.1069	1.2997	0.5082	1.2997	0.5082	1.376	0.2365
													67	0.2632	1.53E-02	0.2374	1.71E-02	0.3504	2.02E-02	1.6941	1.94E-02	1.6747	1.53E-02	1.6941	1.94E-02
68	0.1582	2.50E-02	0.6621	9.57E-02	0.6314	2.21E-02	1.7579	2.85E-02	1.7959	3.30E-02	1.7579	2.85E-02
													69	0.344	1.53E-02	0.3569	4.60E-02	0.4927	1.41E-02	1.8185	2.04E-02	1.825	1.70E-02	1.8185	2.04E-02
70	0.5006	7.65E-02	0.3617	2.18E-02	0.864	1.74E-02	0.9028	1.19E-02	0.8527	6.03E-02	0.9028	1.19E-02
													71	0.1922	0.2744	0.2988	2.77E-02	0.113	2.01E-02	0.793	1.41E-02	0.7759	2.84E-02	0.793	1.41E-02
72	0.2438	3.77E-02	0.1195	3.67E-02	0.407	1.83E-02	0.8333	2.76E-02	0.8495	3.16E-02	0.8333	2.76E-02
													73	0.1389	2.31E-02	0.3036	3.64E-02	0.4909	1.90E-02	1.0368	1.27E-02	1.0336	1.52E-02	1.0368	1.27E-02
74	0.2891	2.34E-02	0.4231	6.45E-02	0.6088	2.75E-02	1.8734	1.32E-02	1.875	1.41E-02	1.8734	1.32E-02
													75	0.2568	3.61E-02	0.1744	3.64E-02	0.6476	2.58E-02	1.2403	1.99E-02	1.2482	2.25E-02	1.2403	1.99E-02
76	0.2616	2.50E-02	0.3052	3.65E-02	0.3408	1.78E-02	1.311	1.34E-02	1.3108	1.56E-02	1.311	1.34E-02
													77	0.281	6.29E-02	0.3736	5.39E-02	0.218	2.05E-02	1.1159	1.61E-02	1.1159	1.20E-02	1.1159	1.61E-02
78	0.2212	4.82E-02	0.1308	4.20E-022	0.3221	1.21E-02	1.6941	1.41E-02	1.6973	1.19E-02	1.6941	1.41E-02
													79	0.2503	6.82E-02	0.2229	2.83E-02	0.415	1.71E-02	1.3076	2.61E-02	1.325	2.22E-02	1.3076	2.61E-02
80	0.2083	5.41E-02	0.2971	1.90E-02	0.3924	3.05E-02	7.69E-02	1.74E-02	8.33E-02	1.32E-02	7.69E-02	1.74E-02
													81	0.4764	7.11E-02	0.415	4.94E-02	0.738	1.33E-02	1.9329	1.84E-02	1.9361	2.06E-02	1.9329	1.84E-02
82	0.3117	4.10E-02	0.3504	4.88E-02	0.3892	2.79E-02	2.36E-02	1.71E-02	2.36E-02	1.71E-02	2.36E-02	1.71E-02
													83	1.2024	0.6534	0.8196	0.5515	0.5733	0.4589	1.0833	0.5242	0.6796	0.4912	1.0651	0.394
84	1.1329	0.4652	0.2542	0.4417	0.331	0.4851	0.8583	0.4417	0.1599	0.6514	0.4501	0.4761

85	0.8718	0.7243	0.2159	0.4037	0.3956	0.4319	0.4757	0.4037	0.8583	0.4319	0.8963	0.4037
													86	0.6554	0.4589	1.357	0.6763	0.7751	0.4652	1.5462	0.6763	0.6558	0.6754	0.8317	0.5389
87	0.0573	0.7192	0.393	0.4423	0.436	0.4003	0.466	0.4423	0.7443	0.6514	0.2011	0.4182
													88	1.2147	0.4911	0.7443	0.6514	1.4118	0.4655	0.1599	0.6518	0.5991	0.4067	0.4118	0.4067
89	0.2507	0.5515	1.5485	0.5697	0.4764	0.5439	0.7697	0.4761	0.59785	0.394	0.2473	0.5537
													90	1.3953	0.5116	1.5008	0.6754	0.3876	0.4067	0.6796	0.4912	1.1251	0.4761	0.7458	0.5515
91	1.1251	0.4761	0.8101	0.5242	1.0403	0.5242	0.667	0.5515	0.4757	0.4037	0.2117	0.5117
													92	0.3554	0.4839	0.4932	0.5828	0.9757	0.551	1.2702	0.5828	1.1019	0.544	0.2804	0.4003
93	0.4931	0.5829	0.2156	0.4037	0.3956	0.4319	0.4757	0.4037	0.8278	0.4067	0.2473	0.5365
													94	0.9956	0.3929	1.2415	0.5579	0.3795	0.5426	1.7898	0.5579	1.0925	0.4308	0.5867	0.4743
95	1.6667	0.5167	0.2158	0.4037	0.6136	0.5476	0.5978	0.4308	0.393	0.4423	0.4118	0.5649
													96	0.9792	0.4951	0.3089	0.4304	0.6298	0.5394	0.5978	0.5838	1.0434	0.544	0.9600	0.4067
97	1.361	0.5951	1.2147	0.4912	0.6621	0.394	0.8904	0.4839	0.689	0.4067	0.4441	0.4081
													98	1.1436	0.5385	0.5991	0.4067	0.7348	0.4912	0.8278	0.4067	1.3781	0.5242	1.5168	0.5579
99	1.0925	0.4308	1.1194	0.5109	0.5025	0.4199	0.6732	0.5109	1.7782	0.551	0.3081	0.432
													100	0.9001	0.5838	1.3953	0.5117	0.6702	0.5364	0.9167	0.5117	1.5484	0.5697	0.8226	0.4516
101	0.9956	0.4513	0.5986	0.6422	0.9366	0.5202	0.1599	0.6422	0.8639	04051	0.45	0.394
													102	0.3962	0.4838	1.1329	0.4653	0.4118	0.3629	0.6822	0.4653	0.8068	0.578	0.589	0.4951

Annotate: experiment pronunciation times N=10.

Table 4, six syllable individual differences analyses of 20 people at random

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20
																					1	000000	000001	001011	001011	101011	001011	001011	100100	111110	001111	001111	101111	101111	101111	101111	101001	100001	111011	111100	111001
2		000000	001010	001010	101110	001011	001111	110001	111011	001111	001111	101111	111111	101111	111111	111000	101000	111110	111001	111000
																					3			000000	001000	001010	001001	101001	111111	111111	101111	001101	101101	111101	101001	111111	111000	101010	111001	111111	111010
4				000000	000010	000001	001001	111111	111111	101111	001111	101111	111111	101001	111111	111000	101000	111001	111111	111000
																					5					000000	000011	101011	111111	111111	101111	001001	101101	111001	101001	111111	111010	100010	101001	110111	111010
6						000000	001011	111011	111111	101111	001110	101111	111111	101110	111111	111011	101011	101010	110011	111011
																					7							000000	011111	111111	001111	101011	111111	111111	111001	011111	111001	111001	001001	110111	111001
8								000000	111011	011000	111111	011111	011111	111111	011000	011111	010111	011111	011000	001111
																					9									000000	111010	111111	011111	011111	011111	110000	010111	111111	111111	011000	011111
10										000000	101111	111100	011100	110100	011011	011111	011111	001111	111011	011111
																					11											000000	111001	111000	111001	111111	111111	111111	101011	111111	111111
12												000000	011011	001011	011111	011101	001111	011101	011111	011111
																					13													000000	111001	011111	011101	011101	011000	001111	010101
14														000000	111111	111001	111011	111010	011101	111011
																					15															000000	011111	011111	011111	011001	011111
16																000000	011001	011011	001111	011000
																					17																	000000	011001	010111	011001
18																		000000	011111	011101
																					19																			000000	011111
20																				000000

Annotate: six bit digital (as: 110011) are represented ba respectively, da, ga, ka, pa, six syllables of ta.Wherein, 0 is that two human world fundamental component phase differential t check at 05 horizontal there was no significant difference: 1 is that two human world fundamental component phase differential t check has conspicuousness in 05 level

Difference.

Except that diagonal line: 000000 (six sounds are completely without difference): 0 pair

000001 (sound is variant): 4 to (comprising: 100000,001000 etc.)

000011 (two sounds are variant) 20 is to (comprising: 110000,101000 etc.)

000111 (three sounds are variant): 30 to (comprising: 111000,101010 etc.)

001111 (four sounds are variant): 54 to (comprising: 111100,110110 etc.)

011111 (five sounds are variant): 53 to (comprising: 111110,1011111 etc.)

111111 (six sounds are variant entirely): 29 pairs

The Figure of description explanation:

(horizontal ordinate is the time to Fig. 1 .Ba syllable sound spectrograph, the s of unit; Ordinate is a frequency, the Hz of unit; Loudness of a sound is represented in brightness, dotted line indication VOT terminating point)

(horizontal ordinate is the time to Fig. 2 .Pa syllable sound spectrograph, the s of unit; Ordinate is a frequency, the Hz of unit; Loudness of a sound is represented in brightness, dotted line indication VOT terminating point)

(horizontal ordinate is a frequency to Fig. 3 .Ba audio frequency spectrogram, the 2.69Hz of unit; Ordinate is a loudness of a sound, the smpl of unit)

(horizontal ordinate is a frequency to Fig. 4 .Pa audio frequency spectrogram, the 2.69Hz of unit; Ordinate is a loudness of a sound, the smpl of unit)

(horizontal ordinate is the time to Fig. 5 .Ba syllable Base Band two pure tone composition sonographs, the s of unit; Ordinate is a loudness of a sound, the smpl of unit)

(horizontal ordinate is the time to Fig. 6 .Pa syllable Base Band two pure tone composition sonographs, the s of unit; Ordinate is a loudness of a sound, the smpl of unit)

(horizontal ordinate is the time to Fig. 7 .Ba syllable Base Band two pure tone composition phase diagrams, the s of unit; Ordinate is a phase place, the π of unit radian, ' π rad ')

(horizontal ordinate is the time to Fig. 8 .Pa syllable Base Band two pure tone composition phase diagrams, the s of unit; Ordinate is a phase place, the π of unit radian, ' π rad ')

(horizontal ordinate is the time to Fig. 9 .Ba syllable Base Band two pure tone composition phase differential figure, the s of unit; Ordinate is a phase differential, the π of unit radian, ' π rad ')

(horizontal ordinate is the time to Figure 10 .Pa syllable Base Band two pure tone composition phase differential figure, the s of unit; Ordinate is a phase differential, the π of unit radian, ' π rad ')

Embodiment

Below in conjunction with embodiment and accompanying drawing the present invention is further detailed:

Key of the present invention is the analysis of fundamental component phase differential and two parameters of VOT of speech syllable.The concrete steps of analyzing are:

1. sound spectrum is handled with VOT and is calculated.Realize the sonograph analysis with the numerical science computational language in the voiceprint analysis system software package.At first read the voice signal of having deposited, add short window and do fast fourier transform FFT conversion, the amplitude of each spectrum component of window center point is represented to form as Fig. 1,2 sound spectrograph with different colours, on screen, carry out the VOT actual measurement.Then to six basic syllables of each speaker (Ba, Da, Ga, Pa, Ta, the VOT of 10 tests Ka) carry out arithmetic mean and calculate, and draw everyone six basic syllable VOT averages and standard deviation.

2. find out fundamental component.Utilizing 1 result, is that the voice signal of 22050Hz adds rectangular window with sampling rate, long 8192 data points of window, and s (i) is i=1..8192 wherein.Then s (i) is done discrete fourier transform and obtain S (i), wherein i=1..8192 generates Fig. 3,4 spectrograms, and the horizontal ordinate of first peak is fundamental frequency among the figure.

3. fundamental component analysis.Get function F S1:FS1 (the i)=S (i) of fundamental frequency point in 2, work as i=b; FS1 (i)=0 is as i ≠ b.Get function F S2:FS2 (i)=S (i) again, work as i=b-3 than low three sampled points of fundamental frequency; FS2 (i)=0 is as i ≠ b-3.FS1 and FS2 are done anti-Fourier transform to be obtained as two pure tone sinusoidal curves among Fig. 5,6.

4. calculating phase differential.Utilize 1,2,3 result, make phase diagram, be i.e. phase differential Fig. 7,8.Get the phase differential of two curves of corresponding VOT value, promptly in Fig. 9,10, get the ordinate value (phase differential) of the corresponding VOT time point of horizontal ordinate.

5. set up database.1. master database: the fundamental frequency twenty percent with 102 people divides phase differential and VOT value to set up two databases respectively; 2. private database: according to the character of using, number, the difference that requires is set up corresponding database, be divided into the judicial person individual assert database, security personnel's person individual assert database, the sound lock person individual assert database, the financial sector person individual assert database and clinical in the person of vocal cords obstacle diagnosis and prognosis is individual assert database etc.

6. identifying.Write down six sounds of Ba, Da, Ga, Ka, Pa, Ta (at least 10 times) of people to be identified, calculating wherein with top program 1,2,3,4,10 times fundamental frequency twenty percent divides phase differential and VOT value and standard deviation thereof, on 0.05 level of difference,, carry out the individual identification of the person with the database schema coupling.

Claims (10)

1. comprise in a plurality of wireless fixed subscriber units (FSUs) system a kind of, each FSU is connected with a user communication device, dispose each FSU and at least one base station communication, this system comprises a plurality of base stations, the base station is connected with a communication network, FSU of a kind of distribution and a certain base station method for communicating, this method comprises:

FSU measures one group of signal of base station intensity;

FSU sends the message of this group signal of base station intensity that measures to a base station;

According to this message and the additional information in being connected to the circuit of this base station, the circuit that is connected to this base station is determined at least one base station that FSU is assigned to, and the base station that wherein has maximum signal is selected;

Be the decision that response has been done by the circuit that is connected to the base station, this base station sends a message that comprises the base station that FSU is assigned to FSU.

2. according to the process of claim 1 wherein that additional information in being connected to the circuit of base station comprises the signal of base station intensity that measures that other each FSU by non-this FSU sends.

3. according to the process of claim 1 wherein that additional information in being connected to the circuit of base station comprises the use pattern of the different base station that FSU can be assigned to.

4. according to the process of claim 1 wherein that additional information in the circuit comprises about the information to the validity of the base station of other FSU.

5. according to the method for claim 1, comprising:

The signal intensity of FSU repeated measurement peripheral base station;

FSU sends the message of measured signal strengths to a base station; With

In response to this message, the updating form of the base station that base station this FSU of transmission can be assigned to.

6. according to the method for claim 1, comprise that FSU communicates by letter with the base station foundation that is assigned to.

7. according to the method for claim 1, comprise the precedence table of the base station that this FSU of FSU maintenance is assigned to.

8. according to the method for claim 7, wherein this preferred list at first based on the signal of base station intensity that measures, upgrades the message that comes from the base station with response then.

9. method according to Claim 8, the message that wherein comes from the base station comprises a new base station preferred list, and substitutes former base station preferred list with the new base station preferred list that comes from this base station.

10. according to the method for claim 7, wherein FSU attempts to finish limit priority base station in the table

Images