JP3328945B2 - Audio encoding device, audio encoding method, and audio decoding method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice codec for digital mobile communication and a voice coding apparatus, a voice coding method and a voice decoding method used in a voice synthesizer for outputting voices of various devices. It is about.

[0002]

2. Description of the Related Art In the field of digital mobile communication,
In order to cope with the increase in the number of subscribers, a speech coding method with a lower bit rate is required, and research and development are being conducted at each research institution. And now the bit rate is 8kb
up to 4.8 kbps.
We are conducting research and development for low bit rates.
However, a speech encoding / decoding device having sufficient performance has not been obtained yet.

[0003] Techniques for realizing a low bit rate are as follows.
Can be divided into different ways. First, as typified by CELP, a first method is a method in which speech in a certain analysis section is encoded by LPC parameters and a driving sound source, and a pitch component is created by a long-term prediction filter. This method is currently the most efficient at medium bit rates of 6.7 kbps to 16 kbps.
However, at 4 kbps or less, a considerable decrease in sound quality is observed, and it is difficult to realize a low bit rate.

A second method is to transmit pitch and one-pitch waveform by performing pitch analysis, utilizing that most of human voice is voiced sound having fundamental frequency. This method is considerably effective in encoding at 6 kbps or less, and can be said to be the most effective method for realizing a low bit rate. However, an error always occurs in pitch detection.
Also, the amount of calculation required for pitch detection is quite large, and it can be said that pitch detection in real time is difficult.

Therefore, it has been considered difficult to realize a low bit rate for the above reasons.

[0006]

As described above, it is difficult to realize a low bit rate with the conventional speech coding / decoding technology. This is because sufficient sound quality cannot be obtained by long-term prediction, and pitch detection is difficult by pitch synchronization.

SUMMARY OF THE INVENTION In view of the above problems, the present invention combines the advantages of the two methods and realizes a low bit rate by using long-term prediction in pitch synchronization coding to prevent pitch errors from occurring. With the goal.

[0008]

To achieve this object, a first aspect of the present invention is an A / D converter for obtaining a digital audio signal from an analog audio signal, and an output of the A / D converter. for the, using past synthesized speech stored in the synthesized waveform storage unit simultaneously a pitch analysis and long-term prediction analysis
A sound analysis unit for obtaining a one-pitch basic waveform and a long-term prediction coefficient, a basic waveform encoding unit for encoding the basic waveform obtained by the acoustic analysis unit, and a basic waveform encoding unit. The obtained basic waveform is decoded, a speech synthesis unit that creates a synthesized waveform from the decoded basic waveform and the first-term long-term prediction coefficient obtained by the acoustic analysis unit, and the speech synthesis unit that creates the synthesized waveform. And a composite waveform storage unit for storing a composite waveform.

According to a second aspect of the present invention, a basic waveform of one pitch and a long-term prediction coefficient are input, the basic waveform is decoded, and a synthesized sound is formed based on the decoded basic waveform and the long-term prediction coefficient. It is configured to

[0010]

According to the above configuration of the present invention, a low bit rate can be realized by encoding a one-pitch basic waveform, and pitch errors can be prevented because long-term prediction is used. Therefore, voice encoding at a low bit rate can be performed efficiently.

[0011]

An embodiment of the present invention will be described below with reference to the drawings.

In FIG. 1, 1 is an A / D converter, 2 is a synthesized waveform storage, 3 is an acoustic analyzer, 4 is a basic waveform encoder, 5 is a speech synthesizer, 6 is a basic waveform decoder, 7 is an audio waveform decoding unit, 8 is a D / A conversion unit, 10 is an encoder, 11
Is a decoder.

Next, the operation of the speech encoding / decoding device according to the first embodiment of the present invention will be described.

First, the function of the encoder 10 will be described with reference to FIG. An A / D converter 1 converts an input voice input from a microphone (not shown) into a digital signal. Next, the acoustic analysis unit 3 captures the audio signal for a certain period of time into a RAM memory (not shown), analyzes the correlation between the analysis section and the synthesized waveform stored in the synthesized waveform storage unit 2, and
The pitch analysis of the analysis section is performed at the same time, and the position of the correlated partial section of the composite waveform and the basic waveform of one pitch are obtained. About this parameter and how to extract the basic waveform,
Details will be described later.

Further, in the basic waveform encoding unit 4,
The basic waveform obtained in the acoustic analysis unit 3 is encoded.
As a specific method, there are a method of performing VQ with a phase matched waveform as it is, and a method of performing encoding after converting to a frequency domain.

Here, the speech synthesizer 5 decodes a one-pitch basic waveform based on the code obtained by the basic waveform encoder 4 and extracts the long-term prediction coefficient extracted by the acoustic analyzer 3 and the basic waveform. And decrypt using
The obtained synthesized waveform is stored in the synthesized waveform storage 2. Since the function of the speech synthesizer 5 is a combination of the functions of the basic waveform decoder 6 and the speech waveform decoder 7 in the decoder 11, the details will be described in the description of the decoder 11.

Here, a method of extracting the parameters and the basic waveform in the acoustic analyzer 3 will be described in detail. The synthesis formula of the present invention is shown in (Equation 1).

[0018]

(Equation 1)

In this (Equation 1), n is the number of the pitch section,
q is a pitch period, β is a long-term prediction coefficient, p is a pitch for long-term prediction, Xnq + i and Xnq + ip are both synthesized waveforms, and Yi is a decoded one-pitch basic waveform.

Therefore, p, q, β, and Yi are determined so that the waveform synthesized by the synthesis formula becomes closest to the original waveform.

Now, if p and q are given, the original waveform

[0022]

(Equation 2)

The error power between the resultant waveform and (1) is as shown in (3) below.

[0024]

(Equation 3)

Here, E is error power, and M is the number of pitch periods in one analysis section. Therefore, when E is the minimum, it is utilized that both the values differentiated by β and Yi become 0. First, by differentiating with β, (Equation 4) is obtained.

[0026]

(Equation 4)

Then, by using the following (Equation 5) and (Equation 6) to simplify the equation and solving for β, (Equation 7) is obtained.

[0028]

(Equation 5)

[0029]

(Equation 6)

[0030]

(Equation 7)

On the other hand, when (Equation 3) is differentiated by Yk, (Equation 8) is obtained.

[0032]

(Equation 8)

Then, (Equation 8) is substituted into (Equation 7) to obtain β
Is determined, and each value Yk is determined using the value. This is performed for all p and q, the error E is evaluated, and p and q with the least error are selected.

However, a full search for p and q requires a large amount of calculation. As a method of reducing the amount of calculation, there are a method of narrowing down and searching for candidates of p and q by the value of Vp, a method of changing q to the value of p when Vp is the maximum, and the like.

Next, the function of the decoder 11 will be described with reference to FIG. First, in the basic waveform decoding unit 6, 1
The pitch basic waveform is synthesized. Then, the speech waveform decoding section 7 uses the one-pitch basic waveform synthesized by the basic waveform decoding section 6 and the long-term prediction coefficient to convert the speech waveform of one analysis section based on the synthesis formula (Equation 1). Combine.
Then, the signal is converted into an analog signal by the D / A converter 8 and output.

In order to verify the effect of the coding of the voice coding / decoding device of the present invention, a preliminary simulation experiment of voice coding / decoding was performed. The voice for evaluation was “explosion sound spread over a plateau in the silver world” uttered by one male, and was encoded at a sampling rate of 8 kHz and 12 bit-PCM. In the simulation, the encoding and
The decoding and the scalar encoding of β were not performed and used as they were. Both p and q are coded with 7 bits. As a result, a segmental S / N ratio of 13.75 dB is obtained,
No pitch error occurred. Also, one pitch waveform is 14
Considering that encoding can be performed at ２０20 dB, encoding can be performed at an S / N ratio of 9 to 12 dB. Therefore, the above object can be achieved.

[0037]

As described above, according to the present invention, pitch errors are prevented from occurring by using long-term prediction in pitch-synchronous coding, so that speech is encoded at a low bit rate without causing pitch errors. And can be decoded.

[Brief description of the drawings]

FIG. 1 is a block diagram of a speech encoding / decoding device according to an embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 A / D conversion part 2 Synthetic waveform storage part 3 Sound analysis part 4 Basic waveform encoding part 5 Speech synthesis part 6 Basic waveform decoding part 7 Speech waveform decoding part 8 D / A conversion part 9 Transmission line 10 Encoder 11 Decoder

──────────────────────────────────────────────────続 き Continuation of front page (56) References JP-A-1-205199 (JP, A) JP-A-62-135899 (JP, A) JP-A-1-126700 (JP, A) JP-A-2- 8900 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G10L 11 / 04,13 / 00 G10L 19/00-19/04

Claims (4)

(57) [Claims] 1. An A / D converter for obtaining a digital audio signal from an analog audio signal, and an output from the A / D converter .
Then , the pitch analysis and the long-term prediction analysis are simultaneously performed using the past synthesized sounds stored in the synthesized waveform storage unit, and 1
A sound analysis unit for obtaining a basic waveform of pitch and a long-term prediction coefficient, a basic waveform encoding unit for encoding the basic waveform obtained by the acoustic analysis unit, and a basic waveform encoding unit obtained by the basic waveform encoding unit. A speech synthesizer for decoding a waveform, a synthesized waveform created from the decoded basic waveform and the long-term prediction coefficient obtained by the acoustic analysis unit, and the synthesized waveform created by the speech synthesis unit are stored. A speech encoding device having the synthesized waveform storage unit. Wherein the input analog audio signal into A / D, against the A / D converted digital audio signal
The pitch using the synthesized waveform before the input
A speech encoding method for simultaneously performing analysis and long-term prediction analysis to obtain a one-pitch basic waveform and a long-term prediction coefficient, wherein a synthesized sound is obtained from the basic waveform and the long-term prediction coefficient, and the synthesized sound is obtained. A speech coding method for obtaining a synthesized speech before the input time by storing the speech. 3. A speech decoding method which receives a one-pitch basic waveform and a long-term prediction coefficient as input, decodes the basic waveform, and forms a synthesized sound based on the decoded basic waveform and the long-term prediction coefficient. 4. The speech decoding method according to claim 3, wherein the synthesized speech is formed by multiplying a past synthesized speech by a long-term prediction coefficient and adding a decoded basic waveform.