JPS63200199A - Speech signal identification system - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a call signal identification method, and in particular to a method for identifying call signals appearing on a telephone line in order to grasp the service status of the telephone line and the telephone usage pattern. This invention relates to a call signal identification method that automatically identifies the type.

[Conventional technology]

Conventionally, this type of call signal identification method focuses on the characteristics of only the active part of the call signal, analyzes the frequency spectrum of the call signal9, analyzes the amount of time fluctuation of the frequency spectrum and the autocorrelation coefficient, and analyzes the results of these analyses. This method identifies signal types by analyzing continuity.

[Problem that the invention seeks to solve]

The conventional call signal identification method described above was based on short-time analysis, so the time required for identification was short. This method has the disadvantage that it is easy to misidentify signals that have relatively similar characteristics, such as human voices and tones (audible signal sounds for telephone exchanges).

An object of the present invention is to solve the above-mentioned drawbacks, analyze the frequency spectrum of a call signal, analyze the time variation of the frequency spectrum and autocorrelation coefficient, and analyze the continuity of these analysis results to characterize the voiced part. It is an object of the present invention to provide a call signal identification method that enables highly reliable identification by analyzing the signal cadence by focusing on the fact that tones turn on and off at regular intervals.

[Means for solving problems]

The call signal identification method of the present invention includes a spectrum analysis unit that performs spectrum analysis as a frequency domain analysis of the call signal to determine maximum power, minimum power, maximum power frequency, bandwidth, and time variation amount of the frequency spectrum; an autocorrelation analysis section that calculates the total power of the signal, an autocorrelation coefficient, and a time variation amount of the autocorrelation coefficient, and detects the active section of the call signal from the outputs of the autocorrelation analysis section and the spectrum analysis section. a cadence analysis section for determining a signal cadence; the spectrum analysis section; and an autocorrelation analysis section.

and an identification condition determining section that determines the signal type of the call signal based on the analysis result obtained from the cadence analyzing section.

[Effect]

In the present invention, the noise level of the line is estimated from the minimum power obtained in spectrum analysis, and the active section is detected by comparing this noise level with the total power. For the section where sound was detected, the frequency characteristics were analyzed from the maximum power frequency and bandwidth, and the temporal characteristics were analyzed from the time variation of the frequency spectrum and the distribution and time variation of the autocorrelation coefficient. Determine the continuity of. Furthermore, we analyze the period of the signal by measuring the time between the active and silent periods, and obtain highly reliable identification results by analyzing the characteristics of the active area of the call signal and analyzing the periodicity of the signal. be able to.

〔Example〕

Next, the present invention will be explained with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention.

This embodiment uses (1) various call signals appearing on the telephone line.
Human voice (voice), (21) audible signal tones (tones) for telephone exchanges such as ringback tones, and (3) data tones (data) such as facsimile signals.
An example of classification is shown in which the spectrum analysis section 100, the cadence analysis section 110. Autocorrelation analysis section 120. An identification condition determination section 130 is provided.

Digitized speech signal 200 is simultaneously input to spectrum analyzer 100 and autocorrelation analyzer 120.

The spectrum analyzer 100 divides the telephone band into eight parts using a bandpass filter (not shown), and calculates the power for each band at each fixed frame period. Next, the maximum power, minimum power, maximum power frequency, bandwidth, and time fluctuation amount of the frequency spectrum are determined from the power for each band. In addition, the autocorrelation analysis unit 120 uses a correlator (not shown) to calculate the zero-order autocorrelation coefficient, which is the total power, by 1 for each fixed frame period.
A first-order autocorrelation coefficient and a second-order autocorrelation coefficient are determined, and then a time variation amount of the autocorrelation coefficient is determined from the first-order and second-order autocorrelation coefficients. The cadence analysis section 110 is the spectrum analysis section 10
0 through the information line 240, and the autocorrelation analysis unit 120 through the information line 250 from the all-type power receiver, and both detect the sound section. Determine the period of the signal by measuring time. The identification condition determination unit 130 uses the maximum power, minimum power, maximum power frequency, bandwidth, and time fluctuation amount of the frequency spectrum obtained from the spectrum analysis unit 100 via the information line 210 and the information line 220 from the cadence analysis unit 110. the time fluctuation amount of the first-order autocorrelation coefficient, second-order autocorrelation coefficient, and autocorrelation coefficient obtained from the autocorrelation analysis unit 120 via the information line 230. Accordingly, voice, tone, and data are identified and an identification result 260 is output. Identification is performed by the spectrum analysis section 100
．． Cadence analysis section 11o.

The information obtained from the autocorrelation analysis unit 120 is audio.

This is performed by determining whether conditions set for each tone and data are satisfied and their continuity. The identification condition determining unit 130 handles the actual data transmitting/receiving unit (Data) and the confirmation signal transmitting/receiving unit (Ack) separately because their signal characteristics are different.
Ack and tones are identified as follows.

(1) Audio... Maximum power frequency: Low range ~
High frequency band, bandwidth: large 9. Time variation of frequency spectrum: large, time variation of autocorrelation coefficient lE2 large, signal period: none.

(21Data...Maximum power frequency: Low to high range, Bandwidth: Large. Time fluctuation amount of 2 frequency spectrum: Large. Time fluctuation amount of autocorrelation coefficient: Small. 1st/2nd order of autocorrelation coefficient. Difference between: Large (However, 1st autocorrelation coefficient: Positive, 2nd autocorrelation coefficient: Negative), Signal period: None.

(31A c k...Maximum power frequency: middle range,
Bandwidth: Small 1. Time variation of frequency spectrum: Small. Time variation of autocorrelation coefficient: Small. Maximum power/minimum power difference: Large. Signal period: None.

(4) Tone: Maximum power frequency: low, bandwidth: small, time variation of 1 frequency spectrum: small, time variation of autocorrelation coefficient: small, maximum power/minimum power difference: large , Signal period: Yes.

〔Effect of the invention〕

As explained above, the present invention obtains highly reliable identification results by capturing the characteristics of the active portion of a call signal through spectrum analysis and autocorrelation analysis, and by performing cadence analysis focusing on the periodicity of the signal. There are effects that can be obtained.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the present invention.

Claims (1)

[Claims] a spectrum analysis unit that performs spectrum analysis to analyze the frequency domain of the call signal to determine the maximum power, minimum power, maximum power frequency, bandwidth, and amount of time fluctuation of the frequency spectrum; and the total power and autocorrelation coefficient of the call signal. and an autocorrelation analysis unit that calculates the amount of time fluctuation of the autocorrelation coefficient, and a cadence that detects the sound section of the call signal from the outputs of the autocorrelation analysis unit and the spectrum analysis unit to determine the signal cadence (period). A call signal identification device comprising: an analysis unit; and an identification condition determination unit that determines the signal type of the call signal based on the analysis results obtained from the spectrum analysis unit, autocorrelation analysis unit, and cadence analysis unit. method.