JPH02273797A - Speaker recognition system - Google Patents

Abstract

PURPOSE:To reduce deterioration in recognition rate with time and to easily perform real-time processing by using the mean PARCOR coefficient of the voice as an input for speaker recognition which uses a neural network. CONSTITUTION:The input voice is divided equally into blocks with time and the voice waveform is analyzed as to partial correlation by blocks of plural sampled values, i.e. at constant intervals of time; and the mean PARCOR coefficient of each block is calculated and inputted to the neural network 20, whose output is inputted to a decision circuit 30 and identified or matched. Consequently, the preprocessing of the input is simple and the time required for the preprocessing is shortened; and the neural network 20 is so structured that the arithmetic processing of the whole network is simple and speedy and the influence of variation due to the period difference of the voice is reduced. Consequently, the deterioration in recognition rate with time is small and the real-time processing is easily performed.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a speaker recognition method suitable for identifying or verifying a speaker from input speech using an online terminal such as an electric lock or an IC card.

[Prior Art] Speaker recognition refers to recognizing a speaker from input speech, and there are two forms: speaker identification and speaker verification.

Speaker identification is to determine which of the registered speakers the input voice belongs to, and speaker verification is to determine whether the input voice can be considered to be the voice of the registered speaker. That's true.

Conventional speaker recognition methods include, for example, the Japanese Patent Publication No. 56-13956.
Publication, Journal of the 5th Society of Electronics and Communication Studies 73/11 Vol. 5
According to the following procedure as described in 6-A No. 11 (Reference 1).

■Extract features related to the speaker included in the input voice.

(2) Calculate the distance between the standard pattern previously extracted in the same manner as (2) above and the feature quantity extracted in (2) above.

■In the case of speaker identification, the speaker of the standard pattern with the minimum distance above is determined to be the speaker of the input voice, and in the case of speaker verification, a threshold value is set in advance for the above distance, and the distance is set to that value. Even if λ is smaller than the input audio (

At this time, in Japanese Patent Publication No. 56-13956 (as mentioned above, the average value and standard deviation of the spectral renewal parameters calculated from the input speech are used as the characteristic quantities) Ru.

[Problems to be Solved by the Invention] However, the above-mentioned conventional one-speaker intelligence method has the following problems.

■As time passes from the time the standard pattern was created, the recognition rate deteriorates. For example (according to the experimental results published in Reference 1), after 3 months have passed since the standard novel turn was created, the speaker identification rate (100%, 00%, 5%) has deteriorated.
The speaker matching rate is inferior to 91% (99%).

■Real-time processing is difficult. ! In conventional speaker recognition methods, it is necessary to use complex feature quantities in order to ensure a certain level of recognition rate or higher, and it is necessary to use complex processing equipment to extract complex feature quantities. This is necessary and requires a lot of processing time.

SUMMARY OF THE INVENTION An object of the present invention is to provide a speaker recognition method that exhibits minimal deterioration in recognition rate over time and that can be easily processed in real time.

[Means for Solving the Problems] The present invention as set forth in claim 1 is a speaker recognition method for recognizing a speaker from input speech using a neural network, wherein an average of the speech is used as input to the neural network. This method uses temporal changes in PARCOR coefficients.

According to a second aspect of the present invention, a temporal change in an average PARCOR coefficient within a certain period of time of audio is used as an input to the neural network.

According to a third aspect of the present invention, the neural network is a hierarchical neural network.

Note that the PARCOR coefficient (partial autocorrelation coefficient) is defined as follows. That is, it is known that there is generally a high proximity correlation between sample values (χ) of audio waveforms.

Therefore, it is assumed that the following linear prediction is possible.

Linear predicted value χ, = -Σα1χ, -1 ... (1)
Linear prediction error ε,=E,~Et'' (2) Here, the sample value of the audio waveform at two times χ, (C
It) ('i=1...., p): (p-th order) linear prediction coefficient Now, in the implementation of the present invention, the linear prediction error ε is 1,
The linear prediction coefficient (α1) is set so that the square mean value of
seek.

Specifically, find (ε)g and calculate its time average (t
t)” is expressed as “a(εt)” / a a r
20r i=1.2. ..., by setting p,
(αi) is obtained from the following equation.

? :,,il! +V1i-1=O+ J=1 + 2
+ ”・* p=・(3) Here, [kn] (n
=1. --1., p) is the (p-order) PARCOR coefficient (partial autocorrelation coefficient), then the PARCOR coefficient k.

, l is defined by the following equation as a normalized correlation coefficient between the forward residual εt1') and the backward residual εt-(n+1)Cb' due to linear prediction.

... (4) Here -ε%f) = χt-Σ (2+Xt-+(αm): Forward prediction coefficient, ε t-(n◆th 2 χ t-(n 1) -Σ
β ノ・χ t−j , (βj): Backward prediction coefficient [Actions] According to the present invention as set forth in claim 1, there are the following effects (1) to (2).

■Deterioration of recognition rate over time is extremely small. This means that
This is confirmed by the experimental results described below,
It is presumed that this is because a neural network can have a structure that is less susceptible to fluctuations due to differences in audio timing.

■As the input to the neural network is the "average FAI'ICOR coefficient of the voice", the preprocessing required to obtain the input is much faster than conventional complex feature extraction.
It is simple, and the time required for this preprocessing is short.

(2) In principle, neural networks allow simple and quick calculation processing of the entire network.

■In principle, each unit that makes up a neural network operates independently, and parallel arithmetic processing is possible. Therefore, calculation processing is quick.

(2) With the above (2) to (4), speaker recognition processing can be easily performed in real time without using a complicated processing device.

Further, according to the present invention as set forth in claim 2, in addition to the above-mentioned effects (1) to (2), there is the following effect (2).

■As the input to the neural network is the average PAR (temporal change in PR coefficient) of the audio within a certain period of time, the processing in the neural network is simple and the time required for this processing is shorter. .

Further, according to the present invention as set forth in claim 3, in addition to the effects (1) to (2) above, there is the following effect (2).

Currently, in a zero-layer neural network,
A simple learning algorithm (pack propagation) as described below has been established, and a neural network that can achieve a high recognition rate can be easily formed.

[Example] Fig. 1 is a schematic diagram showing an example of a speaker recognition system to which the present invention is applied, Fig. 2 is a schematic diagram showing input speech, Fig. 3 is a schematic diagram showing a neural network, and Fig. 4 is a schematic diagram showing a hierarchical neural network, and FIG. 5 is a schematic diagram showing the structure of a unit.

Prior to describing specific embodiments of the present invention, the configuration of the neural network and the learning algorithm will be described.

(1) Due to its structure, neural networks are
The present invention can be roughly divided into two types: the hierarchical network shown in FIG. 3(A) and the interconnected network shown in FIG. 3(B). Hierarchical networks are more useful because simple learning algorithms have been established as described below.

(2) Network structure As shown in Figure 4, the hierarchical network consists of an input layer,
It has a hierarchical structure consisting of an intermediate layer and an output layer.

Each layer is composed of one or more units. The connections are only forward connections such as input layer → middle layer → output layer, and there are no connections within each layer.

(3) Structure of the unit The unit is a model of a neuron in the brain and has a simple structure as shown in FIG. It receives input from other units, sums it up, transforms it using a certain rule (conversion function), and outputs the result. Each connection with another unit is given a variable weight that represents the strength of the connection.

(4) Learning (pack propagation) Network learning is to bring the actual output closer to the target value (desired output). Learn by making changes.

Further, as a learning algorithm, for example, Rumel
hart, D.E., McClelland, J.
L, ”and thePDP Re5earch G
roup, PARALLEL DISTRIBUTE
DPROCESSING, the MIT Pres.
s, 1986. Pack propagation described in (Reference 2) can be used.

Hereinafter, specific examples of the present invention will be described. still,
The recognition system 1 of this embodiment includes a PARCOR analysis unit 1
0, an averaging circuit 15, a neural network 20, and a determination circuit 30 (see FIG. 1).

(A) The learning word was "Tadaima", the input word was "Tadaima", and the time difference between the input word and the learning word was 0 to 3 months.

(B) Speaker ■ Regarding speaker identification, nine registered speakers were identified.

■ Regarding speaker verification, there were 9 registered speakers and 27 learned impostors.
The number of uneducated impostors was 10.

(C) Preprocessing ■ The input voice "Tadaima" is temporally equally divided into four blocks as shown in FIG.

■As shown in Figure 1, the audio waveform is subjected to partial correlation analysis for each block, that is, for each fixed time period, for each of a plurality of (p) sample values (P = 12 in this example), and the Calculate the average PARCOR coefficient.

At this time, the 9th-order PARCOR coefficient, which is the 10 for analysis in the PARCOR analysis section 10, is calculated by the averaging circuit 15 for each block.
is averaged.

(D) Processing and judgment by neural network ■Results of preprocessing (PARCOR analysis unit 10 for each block
The output of the analysis 10 path 15) is input to a three-layer hierarchical neural network 20 as shown in FIG. The input layer 21 has 4 blocks corresponding to p blocks of preprocessing.
It is composed of Xp units (p=12.48 units in this embodiment). The output layer 22 is composed of nine units corresponding to each registered speaker for speaker identification, and two units corresponding to the registered speaker group and other groups for speaker verification.

■The output of the neural network 20 is input to the determination circuit 30, and the unit that outputs the maximum value among the output values of each unit of output N422 is determined by each of the identification recognition section 31 and the verification recognition section 32, and is identified as a result. Or perform verification. However, in implementing the present invention, the output of the neural network 20 is not mechanically subjected to judgment processing such as in the judgment circuit 30;
The determination process may be performed using the human intellect that has obtained the output.

(2) By using the pack propagation of the learning algorithm described above, the network is trained 30,000 times until the output error with respect to the input converges to a constant level, and a network that can guarantee a constant recognition rate is constructed. It should be noted that learning was performed so that the output would be "1" for the unit corresponding to each input speaker and rOJ for the others.

(E) Experiment Using the recognition system 1 described above, speaker identification and speaker matching were experimented.

(a) Speaker identification input speech is ■ “Tadaima” immediately after learning of 9 registered speakers,
■It was named "Tadaima" after 3 months with 9 registered speakers.

As a result, while the identification rate was 100% immediately after learning, the identification rate was 98% after 3 months, and it was observed that the deterioration of the identification rate over time was extremely small.

(b) The input voices for speaker verification are ■ "Tadaima" immediately after learning by 9 registered speakers and 27 learning impostors, ■ "Tadaima" after 3 months by 9 registered speakers and 27 learning impostors, ■ 10 uneducated impostors “
"Tadaima".

As a result, while the matching rate was 100% immediately after learning, the matching rate was 99% after three months, and it was confirmed that there was very little deterioration in the matching rate over time.

(c) Processing speed For the input speech "Tadaima" once, the time required for identification or verification was less than 1 second, and it was confirmed that processing could be performed extremely quickly.

That is, in the recognition system 1, the above (a), (
As shown in the results of b), the recognition rate is high, and the deterioration of the recognition rate over time is extremely small.

Further, in the recognition system 1, as shown in the result (C) above, the speaker recognition process can be performed quickly without using a complicated processing device, and can be easily performed in real time.

In the implementation of the present invention, the "average PARCOR coefficient of audio" is simply used as an input to the neural network, without using temporal changes in the PARCOR coefficient.
may be used.

[Effects of the Invention] As described above, according to the present invention, it is possible to obtain a speaker recognition method in which the deterioration of the recognition rate over time is extremely small and which can be easily processed in real time.

[Brief explanation of drawings]

Fig. 1 is a schematic diagram showing an example of a speaker recognition system to which the present invention is applied, Fig. 2 is a schematic diagram showing input speech, Fig. 3 is a schematic diagram showing a neural network, and Fig. 4 is a schematic diagram showing a hierarchical FIG. 5 is a schematic diagram showing the neural network, and FIG. 5 is a schematic diagram showing the structure of the unit. DESCRIPTION OF SYMBOLS 1... Recognition system, 10... PARCOR analysis part, 15... Averaging circuit, 20... Neural network, 21... Input layer 22... Output layer, 30... Judgment circuit, 31...Identification recognition unit, 32...Verification recognition unit. Patent applicant Sekisui Chemical Co., Ltd. Representative Hiroshi 1) Kaoru Figure 2 Time Figure 3 Figure 4 ↑ Input pattern unit

Claims (3)

[Claims] (1) A speaker recognition method that uses a neural network to recognize a speaker from input speech, and uses an average PARCOR coefficient of the speech as input to the neural network. (2) The speaker recognition method according to claim 1, wherein a temporal change in a typical PARCOR coefficient within a certain time period of speech is used as an input to the neural network. (3) The speaker recognition method according to claim 1 or 2, wherein the neural network is a hierarchical neural network.