JP3342998B2 - Audio decoding method and apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech decoding method and apparatus on the receiving side in speech coded communication, and more particularly, to encoding of frame power (or average amplitude) information of a speech signal as one of parameters. Code Excited Linear Predicti (CELP)
on) Speech coding method or linear predictive coding (LPC: Li
The present invention relates to a receiving-side speech decoding method and apparatus in low-speed (around 4 kbit / s) speech coded communication applied to a wireless link having a low link quality by a near prediction coding (near-prediction coding) method.

[0002]

2. Description of the Related Art With the recent diversification and internationalization of social and economic activities, expectations for mobile communication are rapidly increasing. Above all, the demand for portable mobile phones and cordless phones is growing exponentially. Since such a mobile communication system uses a radio channel, the channel quality is deteriorated due to the effect of fading, and the generation of transmission errors becomes a problem in the digital system. Speech coding technology combined with channel coding (error correction and detection) is indispensable in order to prevent the quality of reproduced voice from deteriorating due to transmission errors.
Considering the effective use of frequency, it is necessary to keep the transmission speed of voice information and redundant information for channel coding low, and the development of efficient voice coding technology and channel coding technology has become an important issue. ing. As an example, in a half-rate speech codec in a digital car telephone system in Japan, 2.15 out of a coding rate of 5.6 kbit / s is used.
kbit / s for channel coding (RCR standard
RCRSTD-27C).

[0003]

[0005] In private communication for public business and general business, a lower transmission rate is required, but at this time, the quality of reproduced voice is not deteriorated as much as possible compared with the car telephone system. Is desirable. For this purpose, a method of lowering the channel coding speed without lowering the voice coding speed can be considered. Here, it is necessary to take measures to avoid quality deterioration due to a decrease in the channel coding speed. As a countermeasure for this, effective use of the parameters in the speech coding method and the statistical properties of the signals of the respective units can be mentioned.

In the following description, CELP will be described as an example. FIG. 1 shows an example of the basic configuration of a CELP encoding apparatus. The encoding apparatus on the transmitting side can be used as it is without changing the conventional encoding apparatus when implementing the present invention. In FIG. 1, a linear prediction analyzer 1 receives a speech signal a1, extracts a spectrum envelope information parameter b1 such as an LSP (line spectrum pair) every frame (40 msec), and outputs the extracted parameter to a synthesis filter 8 and a multiplexer 21. I do. The frame power calculator 2 receives the input audio signal a1, calculates frame power information c1 for each frame (40 msec), and outputs it to the multiplexer 11 and the auditory weighted waveform distortion minimization controller 10.

The adaptive codebook 3 is a codebook for expressing voiced sound source waveforms, and has, for example, 128 types of waveform patterns. The noise codebook 4 is a codebook for expressing an unvoiced sound source waveform, and has, for example, 1024 types of waveform patterns. Adaptive codebook 3 and noise codebook 4
Is the control signal m from the auditory weighted waveform minimization controller 10.
1, codewords d1 and e1 are output. The gain adjuster 5 and the gain adjuster 6 codewords d1 and d1 according to the control signal m1 from the auditory weighted waveform distortion minimizing controller 10.
The gain of e1 is adjusted. The adder 7 adds the code words f1 and g1 whose gains have been adjusted to generate a reproduced excitation signal h1. The reproduced excitation signal h1 is also input to the adaptive codebook 3, and the contents of the codebook are updated.

The synthesis filter 8 adds a spectrum envelope component to the reproduced sound source signal h1 according to the spectrum envelope information parameter b1, and outputs a reproduced sound signal i1. Adder 9
Subtracts the input audio signal a1 from the reproduced audio signal i1 to obtain a prediction error signal j1, and inputs it to the auditory weighted waveform distortion minimization controller 10.

[0007] The auditory weighted waveform distortion minimizing controller 10 comprises:
The adaptive codebook 3 is designed to minimize the perceptually weighted root mean square error of the prediction error signal j1 for each subframe (10 msec).
And the index of the random codebook 4 is controlled and selected by the control signal m1. The auditory weighted waveform distortion minimizing controller 10 uses the frame power information c1 from the frame power calculator 2 to adjust the gain so that the auditory weighted square mean error of the prediction error signal j1 is minimized. The gain of the amplifier 5 and the gain adjuster 6 are controlled and adjusted by the control signal m1. The selected index and gain k1 (indexes of adaptive codebook 3 and noise codebook 4, gains of gain adjuster 5 and gain adjuster 6) are sent to multiplexer 11, and spectral envelope information parameters b1, It is transmitted to the transmission path together with the frame power information c1.

An object of the present invention is to prevent the transmission signal transmitted from the above-mentioned encoding apparatus from deteriorating the quality of reproduced speech caused by a transmission error in a radio channel, and to statistically analyze the parameters of the speech encoding system and the signal of each section. It is an object of the present invention to provide a speech decoding method and apparatus which makes effective use of a property and suppresses deterioration in audibility of reproduced speech.

[0009]

According to a first aspect of the present invention, there is provided a speech decoding method including frame power information of a speech signal.
Code excitation line with predetermined parameters encoded and transmitted
Receiving the transmission signal of the shape predictive coding method,
Audio decoding method for separating and decoding a meter to obtain a reproduced audio signal
, The predetermined parameters received and separated
Of the frame power information and the reproduced audio signal
Comparing the frame power with the frame of the reproduced audio signal.
Power is greater than the frame power information and the difference or
When the ratio exceeds a preset range,
Signal amplitude by a preset level.
It is characterized by having.

[0010] Further, the speech decoding apparatus according to the second aspect of the present invention is configured such that a predetermined signal including frame power information of the speech signal is provided.
Code-excited linear prediction code transmitted with encoded parameters
Receiving the transmission signal of the encoding method, and
Having audio decoding means for separating and decoding to obtain a reproduced audio signal
In the audio decoding device, for each frame of the reproduced audio signal
Playback that calculates the power of the output and outputs the playback audio frame power
An audio frame power calculator, and the reproduced audio frame power
Of the predetermined parameters received and separated
The reproduced audio frame is compared with the frame power information.
Power is greater than the frame power information and the difference
Or when the ratio exceeds a preset range,
The gain information is output, and when it is within the set range, the gain is 1
A gain controller for outputting information; and a gain controller obtained from the gain controller.
Multiplying the reproduced audio signal by the gain information
Gain adjuster for adjusting and outputting the amplitude of the reproduced audio signal
It is characterized in that the bets is provided.

[0011]

FIG. 2 shows a configuration of a CELP decoding apparatus having a transmission error countermeasure function according to the present invention. In the figure,
A portion 33 surrounded by a dotted line is a transmission error countermeasure unit added in the present invention, and the other portions are exactly the same as the basic configuration of the conventional CELP decoding device. In the figure, a separator 21 separates the transmitted audio information parameter a2 and reproduces each parameter. The adaptive codebook 22 and the noise codebook 23 have exactly the same contents as the codebook of the coding apparatus, and output codewords i2 and j2 according to the optimum indexes b2 and e2 transmitted respectively.

The gain controller 24 reproduces the gains g2 and h2 using the gain c2 and the frame power information d2 of the gain adjuster 5 (FIG. 1) and the gain adjuster 6 (FIG. 1). The gain adjusters 25 and 26 adjust the gain of the codewords i2 and j2 using the gains g2 and h2. The adder 27 adds the codewords k2 and r2 whose gains have been adjusted to generate a reproduced excitation signal m2. The reproduced excitation signal m2 is also input to the adaptive codebook 22, and the contents of the codebook are updated. Synthesis filter 28
Adds a spectral envelope component to the reproduced sound source signal m2 according to the spectral envelope information parameter f2, and synthesizes the reproduced sound signal n2. The post filter 29 performs a process for improving the audibility on the reproduced audio signal n2, and outputs a reproduced audio signal s2.

Next, the transmission error countermeasure unit 33 of the present invention will be described. The reproduced audio frame power calculator 30 receives the reproduced audio signal n2 and outputs a reproduced audio frame power q2. The gain controller 31 inputs the frame power d2 from the separator 21 and the reproduced audio frame power q2,
Examine their ratio. Here, if (reproduced audio frame power / frame power)> 1.5, 0.5 is output as the gain t2. In other cases, 1.0 is output as the gain t2. The gain adjuster 32 multiplies the amplitude of the reproduced audio signal s2 by 0.5 when the gain t2 is 0.5, and multiplies the amplitude of the reproduced audio signal s2 by 1.0 when the gain t2 is 1.0. Then, the reproduced signal p2 whose gain has been adjusted is output.

This processing is effective when the error correction code is applied to only parameters that are important for audibility, such as frame power information, for the purpose of reducing the transmission speed. If a transmission error occurs at a high rate in a parameter that is not protected by the error correction code, the amplitude of the reproduced sound becomes abnormally large, and an impulsive sound may be generated. At this time, the frame power of the reproduced sound is protected by the error correction code, and the transmitted frame power is significantly increased. Therefore,
These ratios or differences are examined on a frame basis (or subframes when power information is transmitted on a subframe basis), and when an abnormality is detected, the amplitude of the reproduced signal is attenuated to reduce the strength of the impact sound. Can be suppressed.

FIG. 3 shows an example of the processing waveform of the present invention. FIG. 1A shows an original audio waveform, FIG. 2B shows a reproduced audio waveform by a conventional CELP decoder, and FIG. 1C shows a reproduced audio waveform by a CELP decoder of the present invention. Part T between dotted lines (100 ms
Paying attention to (ec section), in the conventional (B), it can be seen that the amplitude is sharply increased due to a transmission error. This sounds audible as an impact sound. On the other hand, in (C) of the present invention, the amplitude of this portion is attenuated, so that the impact sound is suppressed. However, in this processing example, power information is transmitted in units of subframes (10 msec), and gain adjustment is performed in units of subframes.

Although the components of the circuit of the present invention have been described, the CELP encoder and the CELP decoder are generally realized by a signal processing microprocessor (DSP). Circuit of countermeasure function is also DS
It can be easily realized as a part of the P program.

[0017]

As described in detail above, by implementing the present invention, when an error correction code is applied to only parameters that are important for hearing, such as frame power information, in order to reduce the coding speed, error correction is performed. Since an impulsive sound generated when a transmission error occurs at a high rate in a parameter not protected by a code can be suppressed, there is an extremely great effect that the audible quality of the reproduced sound can be improved.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating an example of a basic configuration of a CELP encoding device.

FIG. 2 is a configuration example diagram of a CELP decoding device having a transmission error countermeasure function of the present invention.

FIG. 3 is a waveform chart showing an example of a processing waveform according to the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Linear prediction analyzer 2 Frame power calculator 3 Adaptive codebook 4 Noise codebook 5,6 Gain adjuster 7,9 Adder 8 Synthesis filter 10 Auditory weighted waveform distortion minimization controller 11 Multiplexer 21 Separator 22 Adaptive codebook 23 Noise codebook 24,31 Gain controller 25,26,32 Gain adjuster 27 Adder 28 Synthesis filter 29 Post filter 30 Reproduced audio power calculator 33 Transmission error countermeasure unit

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-62-36930 (JP, A) JP-A-5-113798 (JP, A) JP-A-5-276122 (JP, A) JP-A-7-76 36496 (JP, A) JP-A-7-58687 (JP, A) JP-A-7-177085 (JP, A) JP-A-7-193519 (JP, A) JP-A-5-100696 (JP, A) (58) Field surveyed (Int.Cl. ⁷ , DB name) G10L 13/00 G10L 19/00-19/02 H03M 7/30 H04B 1/10 H04L 1/00

Claims (2)

(57) [Claims] 1. A predetermined signal including frame power information of an audio signal.
Parameters are transmitted after encoding the parameters.
Receiving the transmission signal of the measurement coding method, and
Audio decoding method to obtain a reproduced audio signal by separating and decoding data
There are, of said predetermined parameters separated by the receiving
The frame power information and the frame power of the reproduced audio signal
And the frame power of the reproduced audio signal is
Larger than the frame power information and the difference or ratio is
When exceeding the set range, the amplitude of the reproduced audio signal is increased.
It is noted that the signal is attenuated by a preset level.
Voice decoding method to be used. 2. A predetermined signal including frame power information of an audio signal.
Parameters are transmitted after encoding the parameters.
Receiving the transmission signal of the measurement coding method, and
Audio decoding means for separating and decoding the data to obtain a reproduced audio signal.
A sound decoding device that calculates the power of each frame of the reproduced sound signal and reproduces the reproduced sound signal.
Reconstructed voice frame power calculator that outputs voice frame power
And the reproduced audio frame power and the received and separated
The frame power information of a predetermined parameter is compared with the frame power information.
In comparison, the reproduced voice frame power is
Range that is larger than the
When exceeded, output gain information smaller than 1 and set
A gain controller that outputs 1 gain information when it is within the range
And reproducing the gain information obtained from the gain controller with the reproduced sound.
Adjust the amplitude of the reproduced audio signal by multiplying the signal
To a gain adjuster to and outputs, characterized in that provided
Audio decoding device.