JPS5912192B2 - audio playback device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an audio reproducing device used in an audio recording/reproducing system, a five-voice response system, etc. using a delta modulation method (including an adaptive delta modulation method).

The delta modulation method basically samples audio signal waveforms, etc. at a cycle set short enough to keep the difference between adjacent sample values within a certain range, and then converts only the magnitude relationship between 10 adjacent sample values into 1-bit data. It is expressed as and quantized.

FIG. 1 is a waveform diagram showing the principle, and the input signal waveform of a is first converted into a staircase waveform as shown in b. This waveform of b is the local 15D/A in the delta modulation encoder.
This is the output waveform of the converter, and the output waveform of the decoder is also the same. And the waveform of b is not shown in c, ``1'',
It is transmitted or stored in memory as a data string of '゛o''. The greatest advantage of this delta modulation method is that it can be encoded at a relatively low bit rate with relatively simple software, but its disadvantage is that it has low fidelity.

One phenomenon with low fidelity is called granular noise, which occurs when the input signal waveform shown in Figure 1a has a small amplitude or
5 is a no-signal portion 11, in which "01-0" alternately occur in the differential data, the output waveform of the local D/A converter b oscillates at the sampling period, and the encoded data c alternately exhibits "0". This is noise caused by the appearance of 1'' and ``00''. The other type is called overslope 30 load, which occurs when the output waveform of the local D/A converter shown in b cannot follow the change in the steeply sloped portion 12 of the input signal waveform shown in Figure 1 a. It is a distortion that causes As a method to compensate for these shortcomings, when the difference data of the input signal waveform 35 is alternately ``1'' and ゜゛O'', quantization in the local D/A converter output waveform is performed to reduce granular noise. Gradually reduce the step width, and if the difference data becomes ``11'' or ``60'' for 3 or 4 or more pieces in a row, conversely, gradually increase the quantization step width to reduce the overslope load. of the method,
Adaptive delta modulation schemes are already in use and have shown considerable improvement.

However, another large distortion occurs in the delta modulation method, including the adaptive type.

This is like a bead between an integer fraction of the sampling frequency and the input signal frequency, and becomes a large noise when the sampling frequency is relatively low. For example, if the sampling frequency is 16kHz and a 1kHz input signal with a steep slope (large level) as shown in Figure 2a arrives, the output waveform of the local D/A converter will be as shown in b, and therefore the encoding The data consists of 8 consecutive 1r゛ and 10'' as shown in c.Here, the input signal frequency is 1kHz.
When the frequency reaches 1.067kHz, which is 16/15 of z, one wave must be expressed with 15 bits.In that case, for example, the output waveform of the local D/A converter is as shown in Figure 3a, and the encoded data is as shown in b. So 01″ and 00″ are 8-J
It continues in such a way that each wave is W bits at a time, and each wave is greatly distorted. Similarly, when the input signal frequency is 1kHz, 1 of 32/31
．． At 0.032kHz, the output waveform and encoded data of the local D/A converter are as shown in Figure 4A and b,
Significant distortion occurs every second wave. These distortions appear as rustling background noise during playback, especially when the level is relatively high, and are extremely annoying. Conventionally, as a countermeasure to this phenomenon, high frequencies have been cut off using an output filter whose cutoff frequency on the high frequency side is relatively low relative to the demodulated waveform.

For example, CVSD (C
OntinuOusVedition1ab1eS10peDelt
A commercially available speech synthesis system using the aM0 (KlratiOn) method with a sampling frequency of 14 kHz has a peak of several DB around 1 kHz, and -18 dB/
An output filter with an attenuation characteristic of 0ct is used.

However, if such an output filter cuts out the high frequency range, the background noise will become less noticeable, but the clarity of the reproduced audio will be considerably impaired, which is not preferable. An object of the present invention is to provide an audio reproduction device that can reduce background noise without impairing clarity when reproducing original audio from data obtained by encoding an audio signal waveform by delta modulation. Generally speaking, speech intelligibility is determined by the phonological beginning part of the speech signal waveform,
In other words, it is greatly influenced by the waveform for a certain period from the beginning of the phoneme, and this tendency is particularly strong in consonant parts.

Furthermore, in the consonant part, relatively high frequency components of around 2 kHz occupy the majority. Therefore, if the waveform at the beginning of a phoneme is obtained with sufficiently high frequency components, the clarity of the reproduced speech will be greatly improved. On the other hand, the distortion explained in FIGS. 3 and 4 mainly occurs in high-level portions as described above, and is less likely to occur in low-level portions such as the beginning of a phoneme. This invention was made with attention to these points, and the output filter provided on the output side of the decoder that demodulates the audio signal waveform from the encoded data has a variable band, so that the output filter remains constant from the start of the phoneme of the encoded data. By moving the cutoff frequency on the high-frequency side of this output filter to a higher side during the period, the output filter normally removes distortion components that cause background noise, and the waveform at the beginning of the phoneme is kept in a sufficiently high frequency range. This was done to improve clarity by passing it all the way to the side. below,
Embodiments of the invention will be described with reference to the drawings. FIG. 5 shows a schematic configuration of an embodiment of the present invention.

The memory 51 stores encoded data obtained by encoding the audio signal waveform by delta modulation (including adaptive delta modulation), and its contents are sequentially read out under control from the controller 52. The output data of this memory 51 is demodulated by a decoder 53 which operates according to the clock signal CK.
This results in a staircase wave as shown in b in FIGS. 1 and 2 or a in FIGS. 3 and 4. The output of the decoder 53 is smoothed by an output filter 54 and then reproduced as audio through an audio amplifier 55 and a speaker 56. The controller 52 has a function of detecting the start point of a phoneme in the output data of the memory 51, and continues for a certain period of time from this point, for example, 1
Only for about 50 ms, the frequency characteristic of the output filter 54 is switched from a characteristic 61 in which the high cutoff frequency is f1 to a characteristic 62 in which the high cutoff frequency is F2, as shown in FIG.

Here, f1 and F2 are, for example, f1=1.2kHz, f2=
The frequency is selected to be approximately 2.4kHz. As a result, the waveform at the beginning of the phoneme, which greatly affects the intelligibility of the output of the decoder 53, passes through the output filter 54 to a sufficiently high frequency, so that the intelligibility of the speech reproduced by the speaker 56 is improved. Furthermore, since the distortion occurring in the high-level region other than the beginning of a phoneme is a high-frequency component, it is almost completely removed by the output filter 54, so that it is not heard as background noise from the speaker 56. FIG. 7 shows a specific example of the configuration of the controller 52.

The address counter 71 specifies the read address of the memory 51 shown in FIG. 5, and increments the address given to the memory 51 by one every time the same clock signal CK as the decoder 52 is applied. This read address is read from the phoneme start address table 73 by the comparator 72.
It is compared with the corresponding data stored in advance in the memory 51, that is, the address information of the data at the beginning of the phoneme in the memory 51. When the comparator 72 detects a match between the two, it outputs one pulse and the table counter 7
By counting up 4, the phoneme start address table 73 prepares the output of data to be compared with the next read address of the memory 51, and at the same time, a monostable multivibrator (mono multi) 75 is activated, and a pulse 76 of a constant width is activated. to occur. The output pulse 76 of the monomulti 75 is given to the output filter 51 as a control signal. On the other hand, the output filter 54 is, for example, as shown in FIG. 8, a secondary low-pass filter consisting of an operational amplifier 81 and an RC circuit 82, with electronic switches 83 and 84 added thereto to switch the resistance of the RC circuit 82. These switches 83 and 84 are the output pulse 76 of the monomulti 75 in FIG.
When the switch is turned on, the cutoff frequency on the high frequency side is switched from f1 to F2.

This invention is not limited to the above embodiments, and can be implemented with various modifications. For example, in the embodiment, the encoded data obtained by encoding the audio signal waveform by delta modulation has been described as being read out from the memory, but the present invention is similarly applicable to data transmitted in an audio transmission system. can do. Furthermore, in the embodiment, one filter is used as the output filter, and its high-frequency cutoff frequency is switched, but two output filters with different high-frequency cutoff frequencies may be switched and used. Of course.

[Brief explanation of the drawing]

Figure 1 is a waveform diagram to explain the principle of the delta modulation method, Figure 2 is a waveform diagram when the input signal frequency is an integer fraction of the sampling frequency, and Figures 3 and 4 are waveform diagrams when the input signal frequency is an integer fraction of the sampling frequency. A waveform diagram when the sampling frequency is not an integer fraction, FIG. 5 is a diagram showing a schematic configuration of an audio reproduction device according to an embodiment of the present invention, and FIG. 6 is a diagram of an output filter for explaining its operation. A characteristic diagram, FIG. 7 is a diagram showing an example of the configuration of the controller in the same embodiment, and FIG. 8 is a diagram showing an example of the configuration of the output filter in the same embodiment. 51...Memory, 52...Controller, 53...Decoder, 54...Output filter, 55...Audio amplifier, 56...・・・
...Speaker.

Claims (1)

[Claims] 1. A decoder that demodulates an audio signal waveform from coded data obtained by encoding the audio signal waveform by delta modulation, an output filter to which the output of this decoder is input, and an output filter that outputs audio from the output of this output filter. It is characterized by comprising means for reproducing, means for detecting a phoneme start point of the encoded data, and control means for moving the high-frequency side cutoff frequency of the output filter upward for a certain period of time from the phoneme start point. Audio playback device. 2 Encoded data is read from memory,
Claim 1, wherein the means for detecting the start point of a phoneme of encoded data detects a point in time when an address in the memory at which data at the start point of a phoneme is stored matches a read address of the memory. audio playback device. 3. The output filter has an electronic switching element for switching the high-side cutoff frequency, and the control means is a monostable type that is activated at the start of a phoneme and generates a control pulse of a constant width to control the switching element. The audio reproduction device according to claim 1, which is a multivibrator.