JPH0431449B2 - - Google Patents

Description

[Detailed Description of the Invention] [Summary] This invention relates to an adaptive differential encoding device, which when a signal is detected by a modem training signal detector,
By switching from differential encoding suitable for audio signals to differential encoding suitable for data, optimal signal transmission is possible regardless of the input signal.

[Industrial application field]

The present invention relates to an improved adaptive differential encoding device for audio signals.

When the above-mentioned adaptive differential encoding device for voice signals is applied to an existing communication network, it is desirable to be able to transmit not only voice signals but also data of voice band modem signals of 9600 bps or less, as well as data at 9600 bps or more.

[Problems to be solved by conventional technology and invention]

Adaptive differential coding method for audio signals (hereinafter referred to as ADPCM)
The coding method (referred to as PCM) utilizes the statistical properties of the power spectrum of audio, an example of which is shown in Figure 3, and attempts to efficiently transmit audio at a lower transmission rate (bandwidth) than conventional PCM. It is.

When applying this ADPCM encoding method to an existing communication network, it is desirable to be able to transmit voice band modem signal data in addition to voice signals, but the power spectrum of the voice band modem signal is As shown in Figure 2, the signal power of the audio signal shows a change of about 40 db, but the fluctuation is small, and the statistical properties are clearly different.

Therefore, for example, as recommended in CCITT,
Specifies the 32kbps ADPCM encoding method
G.721 only compensates for modem signals up to 4800bps, and does not compensate for the transmission of 9600bps modem signals, so there is a problem that it is not possible to transmit modem signals of 9600bps or higher.

[Means for solving problems]

The above problem can be solved by the present invention, in which an adaptive quantizer and an adaptive predictor are connected to an input signal consisting of a voice signal to be adaptively differentially encoded and a voice band modem signal, and the adaptive quantizer and adaptive predictor have encoding characteristics optimized for the voice signal. In a configuration in which the modem training signal detector detects a modem training signal that is first input as a voice band modem signal, a quantizer and a predictor optimized for the voice band modem signal are switched and connected to the input signal. This problem is solved by an adaptive differential encoding device characterized in that the adaptive quantizer and the adaptive predictor are switched and connected again when the output of the modem signal becomes zero.

[Effect]

It is difficult to detect the difference between an audio signal and a modem signal as instantaneous waveforms, so when using a modem, we focus on the training signal for adjusting each part that is sent first, and by detecting the training signal, quantum By switching the quantizer and predictor to the one that is most suitable for the modem, and when the power of the modem signal is gone, returning the quantizer and predictor to the one that is most suitable for the voice signal. , of course, voice band modem signals below 9600 bps
It also enables data transmission of over 9,600 bps.

〔Example〕

FIG. 1 is a block diagram of an ADPCM encoding apparatus according to an embodiment of the present invention, and FIG. 2 is a block diagram of an example modem training signal detector.

In the figure, 1 and 3 are adders, 2 is a quantizer that can be switched between a semi-fixed quantizer and an adaptive quantizer, 4 is a predictor that can be switched between a fixed predictor and an adaptive predictor, and 5
is a modem training signal detector, 6 is a 500Hz bandpass filter, 7 is a 1700Hz bandpass filter, 8 is a 2900Hz bandpass filter, 9 ~
11 is a comparator, 12 is an AND circuit, and 13 is a filter for calculating the mean square.

In Fig. 1, in a normal case, quantizer 2 is an adaptive quantizer with encoding characteristics optimized for audio signals, predictor 4 is also an adaptive predictor, and ADPCM encoding of audio signals is performed. Make it work as a device.

When a modem is used here, when a modem training signal emitted for adjustment of each part is input, it is detected by the modem training signal detector 5 (detailed explanation will be described later), and the quantizer 2 is set to a semi-fixed quantization signal that is optimal for the modem signal. In addition, the predictor 4 is switched to a fixed predictor that is optimal for modem signals, and the encoding device is changed to an encoding device that is optimized for modem signals to encode the modem signal.

It also monitors the power of the modem signal and returns to the original voice mode when this power is gone.

The modem training signal detector 5 will now be explained using FIG. 2.

In the case of Figure 2, the training signal is
We selected A and B patterns, which are simple repeating waveforms specified in CCITT V.29, and this is 500Hz,
Focusing on the fact that it is composed of a composite sine wave with frequencies of 1700Hz and 2900Hz, the input signal is passed through a 500Hz bandpass filter 6, a 1700Hz bandpass filter 7, and a 2900Hz bandpass filter 8, as shown in Figure 2.
The outputs of the comparators 9 to 11 are compared with the threshold values, and the outputs of the comparators 9 to 11 that are at a higher level are ANDed by the AND circuit 12, and the outputs are at the 1 level. When the user gets used to the training signal, the encoding device shown in Fig. 1 is switched to modem mode, while the power of the input signal is detected by the filter 13 that calculates the root mean square, and if the signal power disappears after the training signal is detected, it switches to voice mode. I'm trying to change it back to . In this way, the modem signal can be detected accurately, and when the modem signal is passed, the encoder is switched to modem mode, otherwise it is switched to voice mode. Of course, modem signals in the voice band of 9600 bps or less can also be transmitted.

〔Effect of the invention〕

As explained in detail above, according to the present invention, the ADPCM encoding device for audio signals can process signals of 9600bps or less.
This has the effect of making it possible to transmit modem signals in the voice band of 9600 bps or more.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an ADPCM encoding device according to an embodiment of the present invention, FIG. 2 is a block diagram of an example of a modem training signal detector, and FIG. 3 is a characteristic diagram of an example of voice power spectrum. The figure is a characteristic diagram of the power spectrum of an example modem signal. In the figure, 1 and 3 are adders, 2 is a quantizer that can be switched between a semi-fixed quantizer and an adaptive quantizer,
4 is a predictor that can be switched between a fixed predictor and an adaptive predictor, 5 is a modem training signal detector, and 6 is a
500Hz bandpass filter, 7 is a 1700Hzn bandpass filter, 8 is a 2900Hzn bandpass filter, 9 to 11 are comparators, 12 is an AND circuit,
13 indicates a filter for calculating the root mean square.

Claims (1)

[Claims] 1 Modem training in a configuration in which an adaptive quantizer and an adaptive predictor having encoding characteristics optimized for the voice signal are connected to an input signal consisting of a voice signal to be adaptively differentially encoded and a voice band modem signal. When the signal detector detects a modem training signal that is first input as a voice band modem signal, a quantizer and a predictor optimized for the voice band modem signal are switched and connected to the input signal, and the output of the modem signal is zero. 1. An adaptive differential encoding device characterized in that an adaptive quantizer and an adaptive predictor are switched and connected again when