JPH06124100A - Voice recording and reproducing device - Google Patents

Abstract

PURPOSE:To change the sampling frequency of already recorded data to a value smaller than an original value and to generate a new recording area by providing a function for changing the sampling frequency of the already recorded data. CONSTITUTION:The data is read out from a memory 107 by a memory data readout part 108, and a signal is reproduced as a digital signal by a voice decoding part 109. The sampling frequency of the signal at such a time is still 8kHz by which the data is written in the memory 107 previous time. The signal is inputted in a sampling frequency changing part 114 through a switch 116. The sampling frequency is reduced from 8kHz to 4kHz by the sampling frequency changing part 114, and the data obtained after the change is inputted in a voice encoding part 105 through a switch 115. The data is recorded in the memory 107 as a voice signal whose sampling frequency is 4kHz in the same manner as ordinary recording, and the content of the memory 107 is rewritten.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital voice recording / reproducing apparatus for recording a voice as a digital signal in a memory when recording the voice and reading the signal from the memory for reproduction when reproducing the voice.

[0002]

2. Description of the Related Art A conventional technique is shown in FIG.

Voice is input by input means 301 to 304. Specifically, after passing through the microphone 301, it is amplified by the amplifier 1 of 302, band-limited by the filter 1 of 303, and digitized by the AD converter of 304. The filter 1 of 303 is for cutting the high frequency components in the AD converter of 304 so that aliasing noise is not mixed. For example, when the sampling frequency is 10 kHz, frequency components of 5 kHz or higher are cut. The digitized data is coded by the voice coding unit 305. The encoding here refers to information compression of the data digitized by the AD converter. Various information compression methods have been proposed based on the speech information processing theory.
DM (Adap-tive Delta Modulation) method and ADPC
An example is the M (Adaptive Differential Pulse Code Modulation) method. As an example of specific numerical values in the case of the ADPCM system, audio data digitized by a 10-bit AD converter and expressed as digital data in 1024 stages is information-compressed to 2 to 4 bits. The purpose of information compression is to improve the effect of using a semiconductor memory, which has a much shorter recording time at the same price than a recording medium such as a magnetic tape.

The data encoded and compressed by the voice encoding unit 305 is written in the semiconductor memory 307 by the memory data writing unit 306. This completes the flow of recording.

At the time of reproduction, data is read from the memory by the memory data reading unit 308 and decoded by the voice decoding unit 309. Decoding referred to here is an operation that is the reverse of encoding, and in the above-described example, data of 2 to 4 bits is decoded into 10 bits. The decoded audio data is output by the audio output means 310 to 313.

That is, the DA converter of 310 converts the analog value, the aliasing noise is removed by the filter 2 of 311 and is amplified by the amplifier 2 of 312 and output from the speaker of 313.

[0007]

The conventional digital voice recording / reproducing apparatus has a problem that the sampling frequency is determined at the time of recording and therefore cannot be changed after the recording is completed. As a result, if the memory becomes full, new recording cannot be performed, and it is necessary to erase the recorded data to free the recording area. In addition, for the purpose of efficiently using the memory, it was necessary to set a plurality of sampling frequencies as a recording / playback device and to use the sampling frequencies properly. That is, normally, for example, sampling is performed at 10 kHz (the input band to the AD converter is 5 kHz at this time), recording is performed, and when the memory is low or the minimum necessary sound quality is sufficient,
The idea is to record at half the sampling rate of 5 kHz (the input band to the AD converter at this time is 2.5 kHz). (In the present example, the usage efficiency of the memory is doubled by halving the data amount so that it can be easily understood.) In the above two examples, the sampling frequency is changed after the end of recording. This is a typical example of a problem that arises from the problem that it cannot do things.

[0008]

In the present invention, the sampling frequency of recorded data is provided so that the sampling frequency is determined at the time of recording and cannot be changed after the end of recording. To solve.

[0009]

The above problems are solved by recording once, changing the sampling frequency of the data stored in the memory, and re-storing the data.

[0010]

EXAMPLE FIG. 1 shows an example of the present invention.

The voice is input by the input means 101 to 104. Specifically, after passing through the microphone 101, it is amplified by the amplifier 1 of 102, band-limited by the filter 1 of 103, and digitized by the AD converter of 104. The filter 1 of 103 is for cutting the high frequency component in the AD converter of 104 so that aliasing noise is not mixed. For example, when the sampling frequency is 10 kHz, frequency components of 5 kHz or higher are cut. The digitized data is encoded by the voice encoding unit 105. The encoding here refers to information compression of the data digitized by the AD converter. Various information compression methods have been proposed based on the speech information processing theory.
DM (Adap-tive Delta Modulation) method and ADPC
An example is the M (Adaptive Differential Pulse Code Modulation) method. As an example of specific numerical values in the case of the ADPCM system, audio data digitized by a 10-bit AD converter and expressed as digital data in 1024 stages is information-compressed to 2 to 4 bits. The purpose of information compression is to improve the effect of using a semiconductor memory, which has a much shorter recording time at the same price than a recording medium such as a magnetic tape.

The data coded and compressed by the voice coding unit 105 is written in the semiconductor memory 107 by the memory data writing unit 106. This completes the flow of recording.

At the time of reproduction, data is read from the memory by the memory data reading unit 108 and decoded by the voice decoding unit 109. Decoding referred to here is an operation that is the reverse of encoding, and in the above-described example, data of 2 to 4 bits is decoded into 10 bits. The decoded audio data is output by the audio output means 110 to 113.

That is, the DA converter 110 converts the analog value, the filter 2 111 removes aliasing noise, the amplifier 2 112 amplifies the noise, and the speaker 113 outputs it.

Here, the filters 1 and 2 may share the same one.

With the above, the normal recording / playback is completed.

Now, in the recording / reproducing apparatus of this embodiment, when the memory 107 is full and all the sampling frequencies of the recording data are not at the lower limit of the settable sampling frequency (for example, sampling If the frequency can be set to 4 kHz, 6 kHz, 8 kHz, all recorded data will be 4 kHz.
If not recorded in z. )In other words,
When the sampling frequency of the recorded data can be changed to a lower frequency, the amount of recorded data can be reduced and a new recordable area can be generated in the memory of 107 as follows.

In the following description, the sampling frequency of 8 kHz is changed to 4 kHz. The contents of the description can be realized by software or hardware.

First, the switches 115 and 116 are connected so that the dotted line side is connected. Next, from memory 107,
The data reading unit 08 reads the data, and the voice decoding unit 109 reproduces the signal as a digital voice signal. The sampling frequency of the signal at this time remains 8 kHz at which this data was previously written in the memory 107. 1 through this switch
It is input to the sampling frequency changing unit 14 of FIG. The sampling frequency changer 114 sets the sampling frequency to 8k.
The frequency is reduced from 4 kHz to 4 kHz, and the changed data is input to the speech encoding unit 105 through the switch 115. Here, as an audio signal with a sampling frequency of 4 kHz, data is recorded in the memory 107 and the contents of the memory are rewritten in the same manner as during normal recording. As a result, the amount of data in the memory unit 107 is compressed to half, and a new recording area is generated.

Now, regarding a specific method of constructing the sampling frequency changing unit 114, the following may be adopted in regard to this. First, convert the digital audio signal to LPF (Low
Filter with PassFilter). LPF here
Uses a digital filter that processes a digital signal, unlike the filters 103 and 111. The digital filter is a widely known technology, and many documents have been published. The purpose of filtering is to cut unnecessary frequency components when reducing the sampling frequency. In the above-mentioned embodiment, 8 kHz is reduced to 4 kHz, but in this case, the frequency components of 2 kHz to 4 kHz become unnecessary components. (Generally, when reducing N ₁ kHz to N ₂ kHz, the frequency components from N ₂ kHz / 2 to N ₁ kHz 2 become unnecessary components.) The reason is that if this component exists, the sampling frequency This is because aliasing noise occurs when the frequency is set to 4 kHz.

After removing the aliasing noise component with the digital LPF, the sampling frequency may be reduced by thinning out the digital audio signal. Specifically, set 8 kHz to 4
The case of thinning out to kHz is shown. 8 kHz digital audio signal sequence ..., X (n-3), X (n-2), X (n-
1), X (n), X (n + 1), X (n + 2), X (n
+3), ... Here, X (i) indicates the i-th sampling data. Now, every other new data string ..., X (n-2), X (n), X (n +
2), ... Can be created and used as a 4 kHz digital audio signal sequence. (This is called thinning out 2 to 1.)
FIG. 2 shows the above situation.

The 8 kHz sampling data has a frequency component as shown in (a). The band of voice data is from 0 to 4 kHz, but since it is digital data, the frequency characteristic is such that it is folded back at every 4 kHz as shown in the figure. LP with the data of (a) having the characteristics of (b)
The data is filtered by F to obtain the data of (c). The purpose of filtering is to remove unnecessary shaded areas in the data in (a). Since the band of the data of (c) is from 0 Hz to 2 kHz, even if the sampling frequency is reduced from 8 kHz to 4 kHz and the sampling frequency is reduced from 8 kHz to 4 kHz, it is possible to perform thinning without causing aliasing noise. 8
The 4 kHz data after thinning out the kHz data is shown in (d).

The embodiment in which the sampling frequency is reduced has been described above.

Next, an embodiment in which the sampling frequency is increased will be described.

The configuration of the embodiment in this case is also the configuration of FIG. However, the configuration of the sampling frequency changing unit 114 is different. The following is a description based on FIG. Here, the case where the sampling frequency is increased from 4 kHz to 6 kHz is taken as an example. The 4 kHz data has the frequency component of (a). This is once up-sampled at 12 kHz. The method is such that a 4 kHz sampling data string is ..., X (n-3), X (n-2), X (n-1),
X (n), X (n + 1), X (n + 2), X (n +
3), ..., 0,0, X (n-
3), 0, 0, X (n-2), 0, 0, X (n-1),
0,0, X (n), 0,0, X (n + 1), 0,0, X
Data with a value of 0 may be inserted two by two, such as (n + 2), 0, 0, X (n + 3), 0, 0, .... Generally, the sampling frequency of LkHz is set to MkHz
In order to increase to 0, once, 0 data is inserted into the least common multiple NkHz of L and M by the same method, and upsampling is performed. Now upsampled 12
The kHz sampling data has the frequency component of (b). On the other hand, the data of the characteristic (d) is obtained by filtering with the LPF having the characteristic (c). LPF is band 0 to 2k in the original sampling frequency data (a)
Hz (generally 0 to L / 2 kHz) and the folded component of this, 12 × J ± 2 kHz (generally N × J ± L / 2)
(kHz and J are integers of 1 or more) are set to be removed.

After filtering, the data is thinned out in the same manner as in FIG. 2 to produce the target 6 kHz sampling data. In this case, the sampling frequency is reduced from 12 kHz to 6 kHz so that the sampling frequency is halved, so that the sampling frequency is reduced to 2: 1. (Generally, thinning out to N to M) In this way, the target signal (e) is created. Further, when M is an integer multiple of L, N = M, and therefore it is not necessary to perform decimation after upsampling and filtering with the LPF.

With the above, the increase of the sampling frequency is completed.

[0028]

The recording / playback apparatus of the present invention can be used in the following ways, which has been impossible in the past.

When a new recording is required while the memory is full, a new recording is performed by changing the sampling frequency of the recorded data to a value smaller than the original value to reduce the data amount. Regions can be generated.

In addition, even if the memory is not full, for example, it is possible to erase the recorded contents once listened to, but just in case you want to keep it, the data amount is similarly reduced. By doing so, the memory can be used efficiently.

Further, in the recording / reproducing apparatus of the present invention, when a system configuration is adopted in which the memory is detached from the main body like an IC card and can be connected to another recording / reproducing apparatus, the sampling frequency is reduced except as necessary. Since it is possible to increase the number, the sampling frequencies to be used do not have to match between the recording and reproducing devices. That is, even if the recording / reproducing apparatus A can set sampling frequencies of 4 kHz and 8 kHz, and the recording / reproducing apparatus B can set only the sampling frequency of 6 kHz, according to the present invention, the data recorded by A can be reproduced by B. Is.

[Brief description of drawings]

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

FIG. 2 is a diagram for explaining sampling frequency reduction.

FIG. 3 is a diagram illustrating an increase in sampling frequency.

FIG. 4 is a block diagram of a conventional voice recording / playback apparatus.

[Explanation of symbols]

 101 Microphone 102 Amplifier 1 103 Filter 1 104 AD Converter 105 Speech Encoding Section 106 Memory Data Writing Section 107 Semiconductor Memory 108 Memory Data Reading Section 109 Speech Decoding Section 110 DA Converter 111 Filter 2 112 Amplifier 2 113 Speaker 114 Sampling Frequency Change part 115,116 switch

Claims (3)

[Claims] 1. A means for inputting voice, a means for processing input voice data, a means for writing the processed data in a semiconductor memory, a means for reading the data written in the semiconductor memory, and the above-mentioned means for reading the data. In a voice recording / reproducing apparatus including means for performing reverse processing to reproduce voice data, means for outputting reproduced voice data, and a semiconductor memory connected to these, the sampling frequency of the data written in the semiconductor memory An audio recording / reproducing apparatus characterized in that it is provided with a means for changing the data amount, thereby making it possible to make the data amount variable. 2. A means for inputting voice, a means for processing input voice data, a means for writing the processed data in a semiconductor memory, a means for reading the data written in the semiconductor memory, and the above-mentioned means for reading the data. In a voice recording / reproducing apparatus including means for performing reverse processing to reproduce voice data, means for outputting reproduced voice data, and a semiconductor memory connected to these, the sampling frequency of the data written in the semiconductor memory An audio recording / reproducing device characterized by being provided with a means for reducing the amount of data. 3. A means for inputting voice, a means for processing the input voice data, a means for writing the processed data in a semiconductor memory, a means for reading the data written in the semiconductor memory, and the above-mentioned means for reading the data. A means for reducing the sampling frequency of reproduced data in a sound recording / reproducing apparatus including means for performing reverse processing to reproduce sound data, means for outputting reproduced sound data, and a semiconductor memory connected to these means. The data written in the semiconductor memory is read once by the means to reproduce the audio data, the sampling frequency is reduced by the means for reducing the sampling frequency, and then the audio data having a new sampling frequency is processed by the means. Then, by writing the processed data to the semiconductor memory, A voice recording / playback device characterized by being able to reduce the amount of data in the body memory.