JPS6122370B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION An object of the present invention is to efficiently utilize a recording medium by removing pauses (silent periods) when storing audio signals.

Various band compression methods have been proposed for the purpose of reducing the number of bits as much as possible when digitizing and transmitting audio signals.
Most of them perform compression in the amplitude direction.

Unlike conventional methods, the present invention attempts to reduce the overall number of bits by removing the pause period of the audio signal, and is particularly effective in improving the usage efficiency when using, for example, a magnetic disk as a recording medium. intended to raise.

The fifth is a waveform diagram showing the principle of the present invention. u is the audio signal waveform, and t is the threshold level. Generally, if a state below level t continues for a certain period of time, that period can be regarded as a pause, but if this method is simply used, an unnatural state will occur at both ends of the pause section. Especially when the sound of the S line starts after the pause period ends, it may sound like the sound of the T line with a lisp. To compensate for this drawback, t
The moment from the moment the level reaches below t to the moment when the level becomes higher than t is not regarded as a pose; instead, the pause begins a little after the moment the level reaches t or lower, and the pose starts a little before the moment when the level reaches t or higher. It is sufficient to determine that the process ends. The present invention provides a pose removal method based on this principle.

Next, the method will be explained. In Figure 5,...
Divide the audio signal into equally spaced blocks such as a _-2 , a _-1 , a ₀ , a ₁ , a ₂ , etc. If the level of a subsequent block never exceeds t, and the same holds true for several blocks before and after that block, that block is considered to be a pause. That is, a _ik , a _i-k+1 , ... a _i , a _i
In each block of ₊₁ , _. j and k may be selected appropriately depending on the length of each block, the threshold level t, etc.

FIG. 1 shows the overall configuration of an embodiment to which a recording method according to the present invention is applied. 1 is an audio signal input terminal, 2 is a low-pass filter, 3 is an AD converter, 4 is a pause remover, 5 is a gate, 6 is a buffer memory,
7 is a magnetic disk device, 8 is a write command input terminal,
9 is a write controller, 10 is a buffer memory controller,
11 is a controller for the magnetic disk device, 12 is an audio signal output terminal, 13 is a reduction filter, and 14 is a DA.
Converter, 15 is a gate, 16 is a read command input terminal, and 17 is a continuous output controller.

The operation of this device will be explained. The voice input to the terminal 1 is band-limited by the reduction filter 2, and then PCM encoded by the AD converter 3. The pause remover 4 detects a pause from the PCM code and sends a control signal to the write controller 9, a gate signal to the gate 5,
Gives a delayed signal. When writing to the magnetic disk 7, the buffer memory 6 once writes to the buffer memory 6 at the speed of the input audio signal, and then reads and writes to the magnetic disk 7 at the transfer speed of the disk.
The buffer memory controller 10 performs writing to the buffer memory 6, readout control, address control, etc., and during the pause period, address progress is stopped and no writing is performed. The gate 5 switches between the audio signal and the pause information signal, and multiplexes the audio signal and the pause information signal. The read controller 17 controls the magnetic disk controller 11 and the buffer memory controller 10 based on the read command from the terminal 16.
is controlled in read mode, a pause information signal is detected from the output signal of the buffer memory 6, and during the pause period, the progress of addresses in the buffer memory is stopped, and at the same time, the gate 15 is activated to transmit the signal to the DA converter 14. Make it stop. gate 15
The output is converted into the original analog audio signal by the DA converter 14 and the low-pass filter 13.

Note that the above-mentioned pause information signal is, for example, 8
If we consider the case of encoding with bits, it is possible to express levels from 0 to 255 as sample values, but if we prevent the first level from occurring during the encoding process, we can express that level. The representing code can be used as a pause information signal. In this case 225 is
If it is expressed as 111111111, then every time a pose block appears, the entire block should be replaced with one word 11111111.If one block consists of N words (specimens), then when the pose is Since this only needs to be expressed in one word, there is no need to transmit N-1 words (sample).

FIG. 3 shows a specific configuration example of the pause remover 4, in which 400 to 403 are delay circuits of one block each, 406 to 410 are flip-flops, 405 is a level judge, 412 is a pause judge, and 411 is a pause judge. It is a gate circuit.

The signal coming from the terminal 20 is sent to the level judger 4.
At step 05, it is determined whether or not it is below level t, and the result is latched in flip-flop 406. Flip-flop 406 is reset each time one block of input is completed, and is set when a level greater than or equal to t appears. flipflop 407
.about.410 are each configured so that the state of the previous flip-flop is written every time one block is input. A pause determiner 412 generates a gate signal and a pause information signal for actually removing a signal from the states of the flip-flops 407-410 as a pause section.

FIG. 2 is a time chart illustrating the operation of the embodiment described above. ai indicates the signal of the i-th block. A is an input signal, and B, C, D, and E are delay circuits 400, 401, 402, and 40, respectively.
This is the output signal of No. 3. F ₁ -F ₅ are the outputs of flip-flops 406, 407, 408, and 410, respectively. In the embodiment of FIG. 3, a _i , a _i+1 , a _i+2 ,
This is a circuit that considers a _i to be a pause when both a _i+3 are less than or equal to t. That is, in FIG. 2, if a ₃ to a ₈ are less than or equal to t, then a ₃ , a ₄ , and a ₅ apply to this case, and they are considered to be poses. In Fig. 2, the signal E is the output signal, and G=F ₂
It can be seen that it is sufficient to determine a pause when +F ₃ +F ₄ +F ₅ becomes "L" (low). As another example, when a _i , a _i+1 , a _i+2 , and a _i+3 are all less than or equal to t, when a _i+1 is regarded as a pause, the signal D is used as the output signal, and the G is “L ” can be determined as a pause. In this case, the delay circuit 403 becomes unnecessary, resulting in the configuration shown in FIG. 4. If you want to increase the number of blocks for determining other poses, increase the number of delay circuit flip-flops,
You can achieve your goals using the same way of thinking. In any case, in such a configuration, a pause information signal may be inserted for each block in which G is "L". Pose determiner 412 performs this and generates a pause information signal at terminal 22.

Next, a second embodiment will be explained. This is an attempt to achieve more effective band compression by combining the pause removal method based on the above principle with predictive coding. FIG. 6 is a block diagram showing the overall configuration of the embodiment. In this figure, components indicated by the same numbers as in FIG. 1 perform the same operations as in FIG. 1. 18 is a predictive encoder, and 19 is a predictive decoder.

Figures 7 and 8 show details of the main parts of Figure 6,
The operations are similar to those shown in FIGS. 3 and 4, respectively. The pause determiner 412 also issues the gate signal G, which serves as a gate signal to the gate 411. Therefore, the output of the gate 411 is the output of the AD converter 3 with the pause period replaced with a constant value (in this case "O") (the average value of the output of the AD converter is set to "O"). ). In other words, the signal during the period considered as a pause is approximated by "O". This signal is processed by a predictive encoder 18.
The resulting code is encoded and bit-reduced, and the resulting code is subjected to the above-described pause removal process and recorded on the magnetic disk 7. In other words, pose detection is performed before predictive encoding, and pause removal processing is performed after predictive encoding. This order is important because the poses are defined for the original signal and not for the difference signal.

During playback, gate 1 is
The same signal as the output of the encoder 18 is obtained at the output of the encoder 5 (input of the decoder 19), and a signal converted by the decoder 19 into the original PCM code is obtained. This signal is DA converted by the DA converter 14 and LPF 13
The audio signal is obtained by

9 and 10 show a DPCM encoder and a DPCM decoder, respectively. In FIG. 9, 21 is a PCM code input terminal, and 180 is a subtracter that takes the difference between the input signal and the predicted signal. Reference numeral 181 is a circuit for bit reduction using a compressor (N>n when the input is N bits and the output is n bits). 182 is a circuit for converting a signal whose bits have been reduced by an expander into a signal of the original number of bits; 185 is an output terminal; and 184 is an adder which calculates the difference between the predicted signal and the predicted error. A predictor 183 is a circuit that predicts the value of the next signal appearing at the input terminal 21. In FIG. 10, 190 is an input terminal, 191, 192, and 193 are an expander, an adder, and a predictor, respectively.
It is the same as 84,183. Reference numeral 194 is an output terminal from which the original PCM signal can be obtained.

Even if the compressor, decompressor, predictor, etc. are fixed or adaptively change depending on the state of the input signal, they can be used for the purpose of the present invention.

As is clear from the above embodiments, according to the present invention, when recording audio on a magnetic disk, etc., it is possible to remove the pause period that is included in a considerable proportion of the audio, and to greatly improve the efficiency of use of the recording medium. can be increased to Further, as described in the second embodiment, by combining the predictive encoding methods, the usage efficiency of the recording medium can be further increased.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of an embodiment using the audio signal recording and reproducing method according to the present invention, Fig. 2 is a time chart explaining the principle of the main parts, and Figs. 3 and 4 are detailed configurations of the main parts, respectively. Block diagram showing 5th
The figure is a signal waveform diagram explaining the basic principle of the present invention, FIG. 6 is a block diagram showing another embodiment, and FIGS. 7 and 8 are block diagrams showing the detailed configuration of the main parts of the embodiment. FIGS. 9 and 10 are block diagrams showing detailed configurations of other essential parts of the same embodiment. 4... Pause remover, 7... Magnetic disk device, 17... Read controller.

Claims (1)

[Claims] 1. Means for detecting a pause (silent period) in an audio signal, means for obtaining a signal excluding this pause, means for counting the length of this pause, and means for representing the length of this pause. means for multiplexing a signal and a signal excluding the pause; means for recording the multiplexed signal on a recording medium; and a signal representing the length of the pause and excluding the pause from the signal reproduced from the recording medium. and means for obtaining the original audio signal by inserting a silence signal in accordance with a signal representing the period corresponding to the pause and the length of the pause. A recording and reproducing device for audio signals. 2. A means for PCM encoding an input audio signal, a means for detecting a pause from this PCM signal, a means for predictively encoding the PCM signal, and a signal obtained by excluding the period corresponding to the pause from this predictive code. means for obtaining the length of the pause, means for counting the length of the pause, means for multiplexing a signal representing the length of the pause and a prediction code excluding the pause, and recording the multiplexed signal on a recording medium. means and
means for separating a signal representing the length of the pause from a predicted code excluding the pause from a signal reproduced from the recording medium; 1. An audio signal recording and reproducing device comprising means for obtaining an original audio signal by inserting and decoding the signal.