JP2001285523A - Call discriminating device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a call discriminating device which discriminates between a voice band call and an unlimited digital call, transmitted via a telephone line or an ISDN line using a simple constitution. SOLUTION: This device includes a decoding means A which decodes a service signal, a frequency characteristic detection means B which detects frequency characteristics of the decoded signal, and a call discriminating means C which discriminates whether the service signal is an unlimited digital call or a voice band call, on the basis of detected frequency characteristics. The frequency characteristic detection means B detects the frequency characteristics of the decoded signal in units of frames. The call discrimination means C includes a first discrimination part 11, which discriminates whether each frame is an unlimited digital frame on the basis of frequency characteristics and a second discrimination part 12, which discriminates its being an unlimited digital frame, that the proportion of the number of frames discriminated as being unlimited digital frames is higher than a prescribed value.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a call discriminating apparatus for discriminating a type of a call transmitted through a telephone line or an ISDN line, and more particularly to a call discriminating apparatus for discriminating between an unrestricted digital call and a voice band call. It concerns the device.

[0002]

2. Description of the Related Art In a telephone line or an ISDN line, both voice, voice band data and voice band calls using G3 facsimile and the like, and unrestricted digital calls using video conference equipment and G4 facsimile equipment are transmitted. Since the transmission method and the acoustic characteristics of the speech signal are different between the voice band call and the unrestricted digital call, for example, in a voice signal processing device or a quality measurement device, it may be necessary to handle them differently.

In a voice band call, analog signals such as voice and modem signals are band-limited to a voice band, A / D converted, and transmitted according to a coding rule such as PCM. On the other hand, in an unrestricted digital call, digital information is directly allocated to bits of a communication channel and transmitted, and does not follow a coding rule such as PCM.

In general, telephone line communication equipment manages call type and service type information for each call based on network management information relating to call connection, protocol information from terminals, and the like. For this reason, when it is necessary to distinguish between an unrestricted digital call and a voice band call, the identification is performed by associating with the call to be identified using the management information of these communication facilities.

[0005]

In the conventional system using the management information of the call type and the service type of the communication equipment, a mechanism for exchanging information with the equipment managing such information is required. As a result, there is a problem that the configuration of a device for automatically discriminating between voice band calls and unrestricted digital calls becomes complicated. Also, only the call signal can be obtained,
When management information such as a call type by the communication equipment cannot be used, there is a problem that it is difficult to automatically distinguish a voice band call from an unrestricted digital call.

An object of the present invention is to solve the above-mentioned problems of the prior art and to provide a call discriminating apparatus capable of discriminating a voice band call and an unrestricted digital call transmitted via a telephone line or an ISDN line with a simple configuration. To provide.

[0007]

In order to achieve the above-mentioned object, the present invention provides a call discriminating apparatus for judging the type of a call transmitted through a telephone line or an ISDN line. Decoding means, frequency characteristic detecting means for detecting frequency characteristics of the decoded signal, and, based on the detected frequency characteristics, determining whether the call signal is an unrestricted digital call or a voice band call. And a call discriminating means for discriminating.

In such a feature, if the encoded unrestricted digital call frame and the voice band call frame are decoded and their frequency characteristics are detected, both of them are particularly in the voice band (0.3 to 3.4 kHz). ) Shows specific properties outside. Therefore, by analyzing the respective frequency characteristics, it can be determined whether the call is a frame of an unrestricted digital call or a frame of a voice band call.

[0009]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Before describing the embodiments of the present invention in detail, first, the basic concept of the present invention will be described. In the following, a case where a signal is PCM-coded and transmitted to a telephone line will be described as an example.

In a voice band call, analog signals such as voice and modem signals are A / D-converted and PCM-encoded on the transmitting side and transmitted, and PCM-decoded on the receiving side. At the time of A / D conversion of an analog signal, a band is limited to a voice band (0.3 to 3.4 kHz) in order to suppress the influence of aliasing occurring in the sampling process. For this reason, the frequency characteristics after decoding of the voice band call show unique characteristics in which the power is suppressed in the low and high frequency regions outside the voice band, as shown by the thick solid line in FIG.

On the other hand, in an unrestricted digital call, digital information is directly allocated to bits of a communication channel and transmitted, so that band limitation and PCM encoding are not performed. Therefore, the signal obtained by PCM decoding the unrestricted digital signal has frequency characteristics unrelated to the acoustic characteristics of the original digital information as shown by the thin solid line in FIG. As shown in FIG. 7, the characteristic characteristic having a high power even in a frequency region outside the voice band is exhibited.

Thus, the speech signal of each received call is
If decoded, the frequency characteristics represent the call type. Therefore, in the present invention, the call type is determined by focusing on the specificity of the frequency characteristics described above.

In the power comparison in a frequency band outside the voice band, as shown in FIG. 5, in a transmission system in which the cutoff characteristic of the band limitation in the voice band transmission becomes gentle, as shown in FIG. In such a case, an erroneous determination may be caused by leakage of power out of the voice band.
For this reason, the identification determination accuracy is improved by using a determination criterion based on the autocorrelation of the speech signal.

FIG. 2 is a diagram showing a configuration of a quality measuring device 40 to which the call discriminating device of the present invention is applied, and discriminates whether the type of call is an unrestricted digital call or a voice band call. It includes a call determination unit 41 and a quality measurement unit 42 that performs quality measurement for a voice band call.

The quality measuring section 42 measures the quality of all the call data, and thereafter, the call is
If the call is determined to be an unrestricted digital call by 1, the measurement is stopped, and if the call is determined to be a voice band call, the measurement is continued. Alternatively, the measurement may be selectively started only for a call determined as a voice band call by the call determination unit 41 after the determination.

FIG. 1 is a functional block diagram showing a configuration of a main part of the call discriminating section 41, which detects a decoding block A for decoding received data and a frequency characteristic of the decoded speech signal. A frequency characteristic detection block B and a call determination block C for determining a type of a call based on the detection result of the frequency characteristics are mainly configured. Further, a sound detector 3 for detecting a sound frame, and a correlation determination block D and are attached.

In the decoding block A, a decoder 1 decodes a speech signal transmitted from a telephone line or an ISDN line, and stores it in a buffer 2 as linear data. The linear data stored in the buffer 2 is extracted for each fixed frame size, and various processes are repeatedly executed for each frame. In this embodiment,
The frame size is set to 80 samples (= 10 ms: sampling frequency 8 kHz).

The sound detector 3 reads the linear data x [j] (j = 0 to L-1) of a single frame section from the buffer 2 to determine the presence / absence of sound in the frame and determines the result of the determination. A notification is sent to a frame type determiner (first determiner) 11 and a call type determiner (second determiner) 12. Where x
[J] is the linear data of the j-th sample in the frame, and L is the frame size. As the sound detection method, a known method based on a comparison between the long-term average power and the short-time average power can be used.

In the frequency characteristic detection block B, FF
The T processor 4 performs a fast Fourier transform (FFT) on the frame determined by the sound detector 3 in order to analyze its frequency characteristics. The FFT processor 4 reads out, from the buffer 2, 240 samples of linear data for a total of three frames including one frame before and after the frame to be analyzed as the center. Further, data having a linear value of 0 is uniformly added to both ends so that the total number of samples becomes 256, and then an FFT process is performed by multiplying by a window function such as a Hamming window. Further, based on the following equation 1, the power spectrum Sx in the frame is
[K] is calculated.

[0020]

(Equation 1)

Where k is a frequency sample and Xr
e [k] and Xim [k] are the real and imaginary parts of the FFT processing result at the k-th frequency sample, respectively, and Sx [k] is the power spectrum at the k-th frequency sample.

Out-of-band low-band power calculator 5-1 and out-of-band high-band power calculator 5-2, and first-band power calculator 6-1 to N-th band power calculator 6-N
Then, based on the power spectrum calculated by the FFT processor 4, the power for each frequency band specified in advance is calculated. In this embodiment, as shown in Table 1 and FIG. 6, the frequency region in the audio band is divided into six bands.

[0023]

[Table 1]

Also, in this embodiment, the lower band (up to 0.25 kHz) outside the voice band and the first band (0.2
5 kHz) are allocated so that adjacent and continuous frequency samples are obtained. However, in order to suppress the influence of power leakage into the low-frequency portion outside the voice band due to insufficient band suppression of voice band calls. Alternatively, the band may be allocated so that a gap is generated between the low band outside the audio band and the first band within the audio band.

The out-of-band low-band power calculator 5-1 and the out-of-band high-band power calculator 5-2 use the following equation (2).
According to 3, the respective band powers of the low band and the high band outside the audio band are calculated, and the calculation result is notified to the frame type determination unit 11.

[0026]

(Equation 2)

[0027]

(Equation 3)

Here, PL is the low-frequency power outside the voice band, W
L is the number of frequency samples in the low band outside the audio band, PH is the high band power outside the audio band, WH is the number of frequency samples in the high band outside the audio band, and K0 is the first frequency sample number in the high band outside the audio band. Represent.

The first band average power calculator 6-1 to the Nth
The band average power calculator 6-N calculates the power of each of the six divided bands in the voice band according to the following equation 4, and notifies the maximum value detector 7 of the calculation result.

[0030]

(Equation 4)

Here, P [n] is the power of the n-th band in the voice band, W is the number of frequency samples in the same band, and Kv is the number of the first frequency sample in the voice band. In the maximum value detector 7, the powers from the first band to the Nth band in the audio band are compared with each other, and the maximum band power is notified to the frame type determination unit 11 as the maximum band power Pmax in the audio band. .

In the correlation determination block D, the first-order autocorrelation calculator 8 and the second-order autocorrelation calculator 9 use the linear data of the frame detected by the sound detector 3 to obtain the following equation 5, The normalized first and second order autocorrelation is calculated based on the calculation formula 6, and the calculation result is output to the autocorrelation threshold determiner 10.

[0033]

(Equation 5)

[0034]

(Equation 6)

Here, x [j] is the j-th frame to be processed.
The linear data value in the second sample, L is the frame size, C1 is the primary normalized autocorrelation value, and C2 is the secondary normalized autocorrelation value.

The autocorrelation threshold determiner 10 compares the sum of the absolute values of the primary and secondary normalized autocorrelation values with a predetermined threshold value based on the following equation 7, and The degree of correlation is determined. When the following equation 7 is satisfied, the correlation flag of the frame is determined to be low, and the correlation flag is set to 1. In other cases, the correlation flag is set to 0 and the correlation flag is set to 0.
Notify 1. In this embodiment, the threshold value Thr
Crr is set to 0.5.

[0037]

(Equation 7)

In the call discrimination block C, the frame type discriminator 11 determines the detection result of the sound detector 3 and the detection result 7 of the maximum value detector 7 and each power, as will be described in detail below with reference to a flowchart. It is determined whether each frame is a voice band signal or an unrestricted digital signal based on a comparison result with the calculation results by the calculators 5-1 and 5-2 and a determination result by the autocorrelation threshold determiner 10. . The call type determiner 12 determines, based on the determination result of each frame by the frame type determiner 11, whether the call of the frame is a voice band call or an unrestricted digital call.

FIG. 3 is a flowchart showing the operation of the frame type judging unit 11 and the call type judging unit 12.

In step S1, the frame type determining unit 1
In 1, it is determined whether the sound detector 3 recognizes the frame to be determined as a sound frame, and the determination result is notified to the call type determiner 12. The call type determiner 12
Upon receiving the determination result indicating that the frame is a voice frame, in step S2, the voice frame counter Na is incremented. When the determination result indicating that the frame is a silent frame is received, the process returns to step S1 to determine the next frame.

In step S3, the out-of-band low-band power PL and the out-of-band high-band power PH of the frame determined to have sound are determined by the frame type determiner 11 in relation to the maximum value Pmax by the following equation. It is determined whether any one of 8 and 9 is satisfied.

[0042]

(Equation 8)

[0043]

(Equation 9)

It should be noted that αL and αH are margins of the low band power PL and the high band power PH with respect to the maximum value Pmax. In the present embodiment, αL = αH = 0.5 And

Here, either of the equations 8 and 9 holds, and either of the out-of-band powers PL and PH is not significantly smaller than the maximum value Pmax, in other words, each of the out-of-band powers P
If L and PH are equal to or greater than the maximum value Pmax, the process proceeds to step S4 on the assumption that the frame is an unrestricted digital frame. In step S4, the frame type determiner 11 refers to the correlation flag from the autocorrelation threshold determiner 10. If the value of the correlation flag is 1 (the correlation is low), it is determined that the speech signal of the frame is a truly unrestricted digital signal, and step S
The call type determination unit 12 increments the unrestricted digital frame counter Nd. If the value of the correlation flag is 0 (the correlation is high), the process returns to step S1 to process the next frame.

In step S6, it is determined whether or not a predetermined determination period for determining an unrestricted digital call has ended. This determination period can be set until the determination of the predetermined number of frames is completed or until the call of the determination target speech signal is completed.

When the determination period is over, in step S7, the call type determiner 12 determines the number Na of frames determined to be sound by the sound detector 3 and the frame type determiner 11 outputs the unlimited digital signal. It is determined whether or not the determined number of frames Nd satisfies Expression 10 below. Here, ThrN is a threshold for determining an unrestricted digital call, and in this embodiment, ThrN = 0.
It was set to 9.

[0048]

(Equation 10)

If Equation 10 is satisfied and 90% or more of the sound frames are unrestricted digital frames, step S8
, The call is determined to be an unrestricted digital call.
If the expression 10 is not satisfied, in step S9,
The call is determined to be a voice band call.

In the above-described embodiment, the FFT (Fast Fourier Transform) processor 4 is used to calculate the power for each band in discriminating the unrestricted digital signal. Alternatively, the power for each band may be calculated.

[0051]

According to the present invention, it is possible to determine whether a received call is an unrestricted digital call or a voice band call based on the frequency characteristics of a signal obtained by decoding a speech signal. Therefore, it is possible to perform a call determination based on only a call signal with a simple configuration.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of an embodiment of the present invention.

FIG. 2 is a block diagram of a quality measuring device to which the call discriminating device of the present invention is applied.

FIG. 3 is a flowchart showing the operation of the present invention.

FIG. 4 is a diagram showing frequency characteristics after decoding of an unrestricted digital call and a voice band call.

FIG. 5 is a diagram showing frequency characteristics of a voice band call in which power is concentrated in a low frequency band.

FIG. 6 is a diagram illustrating an example of band division at the time of frequency analysis of an audio band.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Decoder, 2 ... Buffer, 3 ... Sound detector, 4 ... F
FT processor, 5-1 ... Low band power calculator out of voice band, 5
-2: High-band power calculator out of voice band, 6-1 to N: First
~ N band power calculator, 7 ... Maximum value detector, 11 ... Frame type determiner, 12 ... Call type determiner, 40 ... Quality measuring device, 41 ... Call determining unit, 42 ... Quality measuring unit

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) H04Q 1/46 G10L 3/00 551A // G10L 101: 02 7/08 A 101: 14 9/08 301A ( 72) Inventor Minoru Oda 2-1-15 Ohara, Kamifukuoka-shi, Saitama F-term in K.D. Laboratory (reference) 5D015 AA06 CC03 CC04 DD03 HH23 KK01 5K064 AA03 AA07 BA08 BA19 BB12 DB08 DB26 5K101 LL03 PP06 UU11 UU16 9A001 BB06 EE04 HH16 HH34 JJ12 KK56

Claims (6)

[Claims] 1. A call discriminating apparatus for determining a type of a call transmitted via a telephone line or an ISDN line, a decoding means for decoding a call signal, and a frequency for detecting a frequency characteristic of the decoded signal. A call discriminating apparatus comprising: a characteristic detecting unit; and a call discriminating unit that discriminates, based on the detected frequency characteristics, whether the call signal is an unrestricted digital call or a voice band call. 2. The frequency characteristic detecting means detects a frequency characteristic of a decoded signal on a frame-by-frame basis, and the call determining means determines that each frame is an unrestricted digital call based on the detected frequency characteristic. A first discriminating unit for discriminating whether or not the frame is a non-restricted digital call, and a second discriminating unit for discriminating this as an unrestricted digital call when a ratio of the number of frames determined to be a non-restricted digital call is higher than a predetermined value. The call discriminating apparatus according to claim 1, further comprising a discriminating unit. 3. The frequency characteristic detecting means includes: an out-of-band power calculating means for calculating a band power outside a voice band for each frame of a decoded speech signal; and a power of each band in a voice band. A plurality of in-band power calculating means for respectively calculating; and a maximum value detecting means for detecting a maximum value of each band power calculated by each of the in-band power calculating means. 3. The frame according to claim 1, wherein a frame whose power is not significantly smaller than said maximum value is determined as a frame of an unrestricted digital call.
A call discriminating apparatus according to claim 1. 4. The call discriminating apparatus according to claim 3, wherein said out-of-voice-band power calculating means calculates respective low-band and high-band powers outside the voice band. 5. A speech detection means for detecting speech in the decoded signal in frame units, wherein the call discrimination means performs call discrimination only for speech frames. The call discriminating apparatus according to any one of claims 1 to 4, wherein 6. An autocorrelation calculating means for calculating a first-order and a second-order normalized autocorrelation for each frame of a decoded signal, and a comparing means for comparing each of the autocorrelations with a predetermined reference value. The call discriminating means according to any one of claims 1 to 5, wherein the call discriminating means determines a frame whose autocorrelation is lower than a predetermined reference value as a frame of an unrestricted digital call. apparatus.