KR100651731B1 - Apparatus and method for variable frame speech encoding/decoding - Google Patents

Abstract

The present invention classifies an input speech signal according to the characteristics of the input speech signal and performs encoding using a frame size, a quantizer structure, and a bit allocation structure corresponding to the classified classification, or is used by a network state or a call counterpart. The present invention relates to a speech encoding / decoding apparatus capable of adjusting a frame size according to a type of speech encoder and a method thereof. The present invention can improve the performance of the speech encoding apparatus by optimally adjusting the size of the frame, the structure of the quantizer and the bit allocation according to the characteristics of the input speech, and the size of the frame according to the type of speech encoder used by the call counterpart. By adjusting, you can reduce overall call delay.

Description

Apparatus and method for variable frame speech encoding / decoding}

1 is a diagram illustrating a configuration of an embodiment of a speech encoding apparatus and a speech decoding apparatus according to the present invention for optimally performing encoding / decoding according to characteristics of an input speech signal.

2 is a diagram illustrating an example of determining an input speech grade by an input speech rating determiner according to the present invention for optimally performing encoding according to characteristics of an input speech signal.

3 is a diagram illustrating a configuration of a variable adjusted speech encoder of a speech encoder according to an embodiment of the present invention, which performs encoding optimally according to characteristics of an input speech signal.

4 is a diagram illustrating a configuration of a variable adjusted speech decoder of a speech decoding apparatus according to the present invention for optimally performing decoding according to characteristics of an input speech signal.

5A and 5B illustrate a flow of a speech encoding method and a decoding method according to the present invention for optimally performing encoding / decoding according to characteristics of an input speech signal.

FIG. 6 is a diagram illustrating a configuration of an embodiment of a speech encoding / decoding apparatus according to the present invention for reducing a delay required for a voice call in consideration of network conditions. FIG.

7A and 7B are a flowchart illustrating a flow of an embodiment of a speech encoding / decoding method according to the present invention for reducing a delay required for a voice call in consideration of network conditions;

FIG. 8 is a diagram showing the configuration of an embodiment of a speech encoding / decoding apparatus for adjusting the size of a frame according to the type of speech encoder on the receiving side; FIG.

9A and 9B are flowcharts illustrating a flow of an embodiment of a speech encoding / decoding method for adjusting a frame size according to a type of a speech encoder on a receiving side.

The present invention relates to a speech encoding / decoding apparatus and a method thereof, and more particularly, to adjust a frame size, a quantizer structure, and a bit-allocation method according to characteristics of an input speech signal for efficient speech encoding. The present invention also relates to a speech encoding / decoding apparatus and method for adjusting a frame size according to a type of a speech encoder of a call counterpart.

Conventionally, various encoding methods for digitizing and processing a speech signal have been proposed and used. Conventionally, the most widely used coding method is CELP (Waveform coding) such as Pulse Code Modulation (PCM) and CELP (mainstream in the standard of the International Telecommunication Union-Telecommunication Standardization Sector). There is a hybrid coding method in which waveform coding and parameter coding are combined, such as Code_Excited Linear Predicition.

In most hybrid coding methods, for efficient compression of speech signals, spectral information and excitation signals representing the vocal tract transfer function are based on a production model of the speech signals. Separated into components, each is modeled, quantized, and transmitted in an appropriate manner. Representative hybrid coding techniques include G.723.1 and G.729 of ITU-T, and Adaptive Multi-Rate (AMR) encoding method to be used for IMT-2000.

ITU-T G.723.1 is a standardized speech coder for compressing multimedia signals using fewer bits.It is an algorithm that compresses and restores input speech by 30ms frame unit and compresses / restores at two data rates of 5.3 / 6.3 kbit / s. Toll quality.

ITU-T G.729 is an algorithm that divides the input voice into 10ms and compresses and restores it at a bit rate of 8 kbit / s. It provides voice quality of wired networks, and is widely used in the VoIP field along with G.723.1. For the efficient implementation of G.729, which requires a large amount of computation, G.729A, a reduced complexity version while maintaining the frame size and bit-compatibility of G.729, is also widely used.

In addition, there is an adaptive multi-rate coder for next-generation voice communications. Types of this coder include a narrowband adaptive multi-rate (AMR-NB) encoder for processing telephone line band speech and a wideband adaptive multiple-rate (AMR-WB) encoder for processing wideband signals. Both encoders analyze and encode input speech in units of 20ms.

A conventional speech coder extracts and quantizes speech spectrum information and excitation signal information for each frame using a CELP algorithm based on a speech utterance model in a wired / wireless speech communication system. However, since the conventional speech coder using the CELP algorithm uses the same frame size regardless of the characteristics of the input speech, performance degradation occurs in terms of sound quality or encoding efficiency.

In particular, when the size of the frame for parameter analysis, such as G.729, is 10 ms, it is suitable for modeling a rapidly changing transition section, but the coding efficiency is inferior in a static section such as voiced sound.

On the other hand, the 30ms frame size used in G.723.1 is suitable for voiced sound interval coding, but the frequency of transmission of spectral information is not sufficient in the transition interval, resulting in a large distortion of the spectral information in the subframe.

That is, conventional encoders using the same analysis frame size, quantizer structure, and bit-allocation regardless of the characteristics of the input speech have a problem in that the performance variation increases according to the characteristics of the input speech.

In addition, the conventional speech coder always operates at a fixed frame size, and a frame size is determined for each speech coder. For example, G.723.1 has 30msec, G.729 has 10msec and AMR-NB coder has a frame size of 20msec and always processes speech signals with the same frame size.

Recently, there is a growing interest in Voice over IP (VoIP) for transmitting voice data through an IP network. In general, it is known that an end-to-end delay of 150msec or less can provide a good quality service during a voice call. If the delay is long, reverberation occurs and the conversation does not go smoothly. In addition, it is difficult to maintain a constant delay in the packet network because the end-to-end delay can be continuously changed during a call. For high quality voice service, the delay should be less than 150msec and this delay should be maintained during the call.

If the voice coder does not match with the party you want to talk to, you can use transcodec to make the call. In the current packet network, the call cannot be made unless the other party has the same voice coder, but the IP network user using the different voice coders and the subscribers of the wireless network support the call through transcodec.

Currently, CDMA uses many voice coders such as Enhanced Variable Rate Coder (EVRC) and Qualcomm Code Excited Linear Prediction (QCELP), and voice coders such as G.729 and G.723.1 are mainly used in VoIP. For example, an IP phone user using G.723.1 needs a Transcodec to make a call to a wireless network subscriber using EVRC.

Transcodec converts the bitstreams encoded and encoded by G.723.1 into bitstreams that can be decoded by EVRC, and converts the bitstreams encoded and encoded by EVRC to be decoded by the G.723.1 decoder. Transcoding basically requires a delay of the maximum common multiple of the frame size of the speech encoder used by both users.

Therefore, 60msec delay is basically required for transcoding for the call between G.723.1 and the subscription using the EVRC encoder. This increase in delay can affect the overall quality of service.

The technical problem to be achieved by the present invention is to variably adjust the frame size according to the characteristics of the input speech, to adaptively use the structure of the quantizer for quantizing the spectral information and the excitation signal, and to quantize the spectral information and the excitation signal. There is provided a speech encoding / decoding apparatus and a method for adaptively adjusting a combination of bits allocated to the to improve the performance of speech encoding / decoding.

Another technical problem to be solved by the present invention is to adjust the frame size of the voice encoder and the number of frames per packet according to the state of the network or the type of encoder of the call partner in the packet network to adjust the overall delay required for transmitting the voice data or to perform the transcoding. The present invention provides a speech encoding / decoding apparatus and method for improving service quality by reducing delay.

Another object of the present invention is to provide a speech encoding / decoding apparatus and method for differentiating a frame size for packet transmission and a frame size for packet encoding.

In accordance with one aspect of the present invention, there is provided an apparatus for encoding a speech, comprising: an input speech rating determiner classifying a rating of an input speech into a transition section and a static section; A variable adjusted speech encoder that variably performs encoding using a different frame size, quantizer structure, and bit-allocation according to the determined grade; And a multiplexer for outputting a bit string of the input speech encoded with the variable frame size.

In order to achieve the above technical problem, an embodiment of the speech encoding method according to the present invention comprises: (a) classifying a grade of an input speech into a transition period and a static period; (b) variably performing encoding using a different frame size, quantizer structure, and bit-allocation according to the determined class; And (c) outputting a bit string of the input speech encoded with the variable frame size.

In order to achieve the above technical problem, an embodiment of a speech decoding apparatus according to the present invention may include receiving a bit string encoded using a different frame size, quantizer structure, and bit-allocation according to a grade of an input speech. A demultiplexer for extracting parameter information necessary for decoding from the bit string; A variable adjusted speech decoder provided for each grade of the input speech and variably performing decoding corresponding to the received grade of the input speech; And a temporary storage unit for temporarily storing the decoded input voice to be continuously output.

In order to achieve the above technical problem, an embodiment of a speech decoding method according to the present invention includes: (a) encoding a bit string encoded using a different frame size, quantizer structure, and bit-allocation according to a grade of an input speech. Extracting parameter information necessary for decoding from the bit string when received; (b) variably performing decoding corresponding to the grade of the received input speech, provided for each grade of the input speech; And (c) temporarily storing the decoded input speech to be continuously output.

Another embodiment of the speech encoding apparatus according to the present invention for achieving the above technical problem is to determine the frame size and the number of frames per packet for the input voice transmission based on the delay information of the network or information on the call counterpart encoder A frame determination unit; A variable adjusted speech encoder that variably encodes the input speech corresponding to the determined frame size and the number of frames; And a multiplexer for outputting a bit string of the input speech encoded with the variable frame size.

Another embodiment of the speech encoding method according to the present invention for achieving the above technical problem, (a) the frame size and packet for coding the speech signal based on the delay information of the network or information on the encoder of the call counterpart Determining the number of frames per frame; (b) variably encoding the speech signal corresponding to the determined frame size and the number of frames; And (c) outputting a bit string of the speech signal encoded in the variable frame size.

Another embodiment of the speech decoding apparatus according to the present invention for achieving the above technical problem is, demultiplexing to extract the parameter information necessary for decoding from the bit string when receiving the encoded bit string based on the delay information of the network part; A variable-adjusted speech decoder provided for each frame size to variably perform decoding corresponding to the size of the received input speech; And a temporary storage unit for temporarily storing the decoded input voice to be continuously output.

Another embodiment of the speech decoding method according to the present invention for achieving the above technical problem, (a) when receiving the coded bit strings based on the delay information of the network to extract the parameter information necessary for decoding from the bit strings Doing; (b) variably performing decoding corresponding to the size of the received input speech provided for each frame size; And (c) temporarily storing the decoded input speech to be continuously output.

Another embodiment of the speech encoding apparatus according to the present invention for achieving the above technical problem, frame size for encoding based on any one of the characteristics of the input speech, the delay information of the network and the codec information of the encoder of the call counterpart. A variable encoder configured to determine and to encode the input speech based on the determined frame size; And a frame transmitter configured to transmit the encoded frame at a constant transmission interval.

Another embodiment of the speech encoding method according to the present invention for achieving the above technical problem, frame size for encoding based on any one of the characteristics of the input speech, network delay information, the codec information of the encoder of the call partner. Determining and encoding the input speech based on the determined frame size; And transmitting the encoded frame at a constant transmission interval.

Accordingly, the frame size, the quantizer structure and the bit allocation are optimally adjusted according to the characteristics of the input voice, and the frame size is adjusted based on the state of the network and the type of the encoder of the call counterpart. It can improve performance.

Hereinafter, a speech encoding / decoding apparatus and a method thereof according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram illustrating a configuration of an embodiment of a speech encoding apparatus and a speech decoding apparatus according to the present invention for optimally performing encoding / decoding according to characteristics of an input speech signal.

Referring to FIG. 1, a voice communication system using a voice encoding apparatus as a transmitter 100 and a voice decoding apparatus as a receiver 150 is illustrated.

The speech encoding apparatus, which is the transmitter 100, is composed of an input speech grade determiner 105, a variable adjusted speech encoder 110, and a multiplexer 115. The speech decoding apparatus as the receiver 150 includes a demultiplexer 155 and a variable adjusted speech decoder 160.

The input speech grade determining unit 105 determines the grade of the input speech. The input speech is classified into static segments, such as transition segments and voiced sound segments, in which the change of the speech signal is large. Since the input voices in the transition section and the static section have different characteristics, it is efficient to use G.729 in the transition section and G.723.1 in the static section. In this way, since the optimum encoding method according to the characteristics of the input speech is different, the input speech rating determiner 105 classifies the grades according to the characteristics of the input speech in order to select the optimal encoding method. The input voice rating determiner 105 may classify the input voice into various grades according to the characteristics of the input voice in addition to the above-described grades of the transition section and the static section.

The input speech rating determiner 105 uses an open-circuit rating method or a closed-circuit rating method to classify the ratings according to the characteristics of the input speech. While the open circuit grading method directly determines the rating according to the characteristics of the input speech, the closed loop grading method determines the rating of the input speech through a feedback process.

The variable adjusted speech encoder 110 encodes the input speech using a frame size, a quantizer structure, and bit-allocation preset in accordance with the grade determined by the input speech grade determiner 105.

The multiplexer 115 considers that the variable adjusted speech encoder 110 uses a variable frame size, and outputs a bit string of the input speech encoded from the variable adjusted speech encoder 110.

The demultiplexer 155 of the receiver 150 receives a bit string output from the multiplexer 115 of the transmitter 100, and extracts parameter information necessary for decoding from the received bit string. The demultiplexer 150 transmits the received input speech bit string to the variable adjusting speech decoder 160 to perform decoding corresponding to the class of the received input speech bit string.

The variable adjustment speech decoder 160 restores the speech signal from the received bit string through a decoder corresponding to the speech class.

2 is a diagram illustrating an example of determining an input speech grade by an input speech rating determiner according to the present invention, which performs encoding optimally according to characteristics of an input speech signal.

The voice signal has various characteristics, and the input voice grade determiner determines which grade the input voice belongs to among the pre-classified classes. Different coding schemes are applied according to the class determined by the input speech class determiner.

3 is a diagram illustrating a configuration of a variable adjusted speech encoder of the speech encoder according to the present invention, which performs encoding optimally according to characteristics of an input speech signal.

Referring to FIG. 3, the variable adjusted speech encoder 110 includes an input speech temporary storage unit 300 and at least one variable speech encoder 305 to 315.

The input voice temporary storage unit 300 stores voice samples in units of frames according to the grade of the input voice determined by the input voice grade determiner 105. The input voice samples stored by the input voice temporary storage unit 300 are transmitted to the variable voice encoders 305 to 315 having the corresponding grades according to the grades of the input voices.

The variable speech encoders 305 to 315 are provided in one-to-one correspondence with the grade determined by the input speech grade determiner 105 and perform encoding according to each grade.

For example, it is assumed that the input speech rating determiner 105 divides the input speech into two grades, a transition section and a static section. Then, the variable speech encoders 305 to 315 exist for encoding the input speech belonging to the transition section and for encoding the input speech belonging to the static section. The input voice rating determiner 305 to 315 determines whether the input voice is a transition interval grade or a static interval grade, and transmits the input voice to the variable voice encoders 305 to 315 according to each grade.

Each of the variable speech encoders 305 to 315 has different frame sizes, different structures of quantizers, and bit-allocation schemes, respectively. Accordingly, the variable adjusted speech encoder 110 may encode the input speech using an optimal encoding method according to each class.

4 is a diagram illustrating a configuration of a variable adjusted speech decoder of a speech decoding apparatus according to the present invention for optimally performing decoding according to characteristics of an input speech signal.

Referring to FIG. 4, the variable adjusted voice decoder 160 includes at least one variable voice decoder 400 to 410 and an output voice temporary storage 415.

When the demultiplexer 155 of the receiver 150 receives an input speech bit string encoded differently according to a grade from the transmitter 100, the variable speech corresponding to the received input speech bit string according to the grade of the input speech bit string Transfer to the decoder 400 to 410.

The variable speech decoders 400 to 410 perform decoding based on the grade of the encoded input speech bit string. The variable voice decoders 400 to 410 of the receiver 150 and the variable voice encoders 305 to 315 of the transmitter 100 correspond one-to-one to perform encoding and decoding, respectively, according to the grade of the input voice.

 The output voice temporary storage unit 415 temporarily stores and outputs the input voice decoded by the variable voice decoders 400 to 410 to enable continuous voice output. That is, since the frame size of the input voice decoded by each of the variable voice decoders 400 to 410 is variable, the output voice temporary storage unit 415 temporarily stores the decoded input voice and outputs them continuously.

5A and 5B are diagrams illustrating a flow of a speech encoding method and a decoding method according to the present invention for optimally performing encoding / decoding according to characteristics of an input speech signal.

Referring to FIG. 5A, the input voice rating determiner 105 determines a rating according to the characteristics of the input voice (S500). The grade according to the characteristics of the input speech is predetermined and there is an encoding scheme optimized for each grade.

The variable adjusted speech encoder 110 encodes and outputs an input speech using a frame size, a quantizer structure, and bit-allocation according to the grade of the input speech (S510).

Referring to FIG. 5B, the demultiplexer 155 receives a bit string of an input speech encoded according to a grade, and the variable speech of the variable-adjusted speech decoder 160 corresponding to a grade of a bit string of the received input speech. Transmission to the decoders 400 to 410.

The variable voice decoders 400 to 410 decode the input voice bit string and output the continuous voice.

1 to 5b relate to a structure of a speech coder that varies frame size and bit allocation according to characteristics of an input signal, and relates to a speech encoding / decoding apparatus having a structure in which a frame size can be changed during a call. .

In addition, the encoding / decoding apparatus and method of the present invention prevents the delay that occurs when the frame sizes are different between callers by setting the frame size according to the voice coder used by the other party as well as during a voice call. .

For example, when making a call from A to B, if the frame size of the speech coder used by B is 20 msec, set the frame size of the speech coder to 20 msec. If the frame size of the speech coder B used is 10 msec. In A, the frame size of the speech encoder is set to 10 msec.

If the frame sizes of the speech coders used in A and B are the same, there is an advantage in Tandem delay. If the frame size of the speech coder used by A is 20 msec and the frame size of the speech coder used by B is 30 msec, a delay of 60 msec is required for A and B to communicate. However, if the frame size is set to 30msec in A, the delay required for the two parties is 30msec.

Therefore, the use of a speech coder having a structure that can set the frame size at the time of a call connection with the frame size of the voice coder used on the other side has an advantage in delaying the tandem time.

Hereinafter, an encoding / decoding apparatus and method for reducing a delay required for a voice call will be described in detail with reference to FIGS. 6 and 9.

FIG. 6 is a diagram illustrating a configuration of an embodiment of a speech encoding / decoding apparatus according to the present invention, which reduces a delay required for a voice call.

Referring to FIG. 6, a voice communication system using a voice encoding apparatus as a transmitter 600 and a voice decoding apparatus as a receiver 650 is illustrated.

The speech encoding apparatus, which is the transmitter 600, is composed of a frame determiner 605, a variable adjusted speech encoder 610, and a multiplexer 615. The speech decoding apparatus, which is the receiver 650, includes a demultiplexer 655 and a variable adjusted speech decoder 660.

The frame determiner 605 determines the frame size for speech encoding and the number of frames per packet. The size of the frame and the number of frames per packet are determined by the state of the network. For example, when the delay of the entire network is increased and the quality of service occurs, the overall delay is reduced by reducing the frame size and the number of frames per packet of the speech encoding apparatus. In addition, when the overall delay of the network is reduced, the size of the frame and the number of frames per packet are increased.

The overall delay can change over the course of a call, so that the size of the frame and the number of frames per packet are constantly adjusted to maintain a constant level of latency, depending on the state of the network during the call.

The variable adjusted speech encoder 610 encodes the input speech signal according to the frame size determined by the frame determiner 605. Since the frame size can be continuously changed during a call, the variable-adjusted voice encoder 610 changes the frame size even during a call so that the sound quality is not degraded.

The multiplexer 615 outputs a bit string of the input speech coded from the variable adjusted speech encoder 610 in consideration of the variable adjusted speech encoder 610 using a variable frame size.

The frame determiner 605 and the input speech grade determiner 105 illustrated in FIG. 1 may be implemented in one configuration to determine an input speech grade and a frame size. In addition, the variable adjusted speech encoder 610 may implement the same function and configuration as the variable adjusted speech encoder of FIG. 1. However, the variable adjusted speech encoder 110 of FIG. 1 performs encoding in response to the speech grade, and the variable adjusted speech encoder 610 of FIG. 6 performs encoding corresponding to the frame size. The multiplexer 615 may implement the same function and configuration as the multiplexer 115 of FIG. 1.

Accordingly, the voice encoding apparatus 600 illustrated in FIG. 6 may be implemented using the speech encoding apparatus 100 according to the present invention illustrated in FIG. 1, and each of the speech encoding apparatuses illustrated in FIGS. 1 and 6 may be implemented. It can be implemented by integrating a function into one coding apparatus.

The demultiplexer 655 of the receiver 650 receives the bit string output from the multiplexer 615 of the transmitter 600. The demultiplexer 655 extracts a parameter necessary for decoding from the received bit string and transfers the parameter to the variable adjusted speech decoder 660. The variable adjusted speech decoder 660 decodes the received bit string. The temporary storage unit (not shown) temporarily outputs the decoded input voice so that the decoded input voice can be continuously output.

The receiver 650 of FIG. 6 may be implemented by the receiver 150 illustrated in FIG. 1 and vice versa, and may be implemented by integrating the receiver illustrated in FIGS. 6 and 1 into one receiver.

7A and 7B are flow charts illustrating the flow of an embodiment of a speech encoding / decoding method according to the present invention for reducing the delay required for a voice call.

Referring to FIG. 7A, the frame determiner 605 determines the frame size and the number of frames per packet according to delay information of the network (S700 and S710). The variable adjustment speech encoder 610 encodes and outputs an input speech signal using the determined frame size (S720 and S730).

Referring to FIG. 7B, the demultiplexer 655 receives a bit string of an encoded input speech (S750), extracts a parameter necessary for decoding from the received bit string, and then decodes the received bit string. In operation 750, the process transmits to S660. The variable adjusted speech decoder 660 variably performs decoding corresponding to the size of the received input speech and outputs it (S760). The temporary storage unit temporarily stores the decoded input voice to be continuously output.

FIG. 8 is a diagram illustrating a configuration of an embodiment of a speech encoding / decoding apparatus that adjusts a frame size according to a type of speech encoder of a receiving side.

Referring to FIG. 8, a speech encoding apparatus, which is a transmitter 800, includes a frame size adaptive speech encoder 805 and a multiplexer 810. The speech decoding apparatus as the receiver 850 includes a demultiplexer 855 and a variable frame size adaptive speech decoder 860.

Transcodec is required for calls between users who have different voice coders. In this case, the delay required for transcoding can be reduced by adjusting the frame size of the speech encoder. In other words, Transcodec is required to make a call between an IP telephone user using a different voice coder and a subscriber of a wireless network. Transcoding requires a delay of at least the largest common multiple of the frame sizes of the encoders used by both sides.

For example, transcoding is required for calls between G.723.1 and users with EVRC, but the delay is at least 60 msec. Therefore, when transcoding is required, if the frame sizes of the speech coders used by both sides are the same, the delay required for transcoding is reduced accordingly. Therefore, the delay required for transcoding can be reduced by adjusting the frame size of the speech coder to be the same as the frame size of the other speech coder.

The frame size adaptive speech encoder 805 encodes an input speech signal with a frame size determined according to the type of speech encoder of the counterpart. The frame size is determined by the type of the other party's voice coder when the call is connected and does not change during the call. The multiplexer 810 outputs a bit string of the input speech encoded by the frame size adaptive speech encoder.

The demultiplexer 855 of the receiver receives a bit string output from the multiplexer 810 of the transmitter. The demultiplexer 855 extracts a parameter for decoding from the received bit string and transfers the parameter to the frame size adaptive speech decoder 860. The frame size adaptive speech encoding / decoding apparatus 800, 850 encodes / decodes a speech signal using a speech signal analysis and quantization table according to the frame size.

9A and 9B are flowcharts illustrating a flow of an embodiment of a speech encoding / decoding method for adjusting a frame size according to a type of a speech encoder on a receiving side.

Referring to FIG. 9A, the frame size adaptive speech encoder 805 encodes a speech signal at a frame size determined according to the type of speech encoder of a counterpart to make a call using transcoding (S900 and S910). The multiplexer 810 outputs a bit string of an input speech encoded with a variable frame size (S920).

Referring to FIG. 9B, the demultiplexer 855 receives a bit stream of an encoded input speech (S950) and transfers the received bit stream to the frame size adaptive speech decoder 860 of the speech decoding apparatus 850. send. The frame size adaptive speech decoder 860 decodes the received bit string (S960), and the temporary storage unit (not shown) temporarily stores the decoded input speech to be continuously output (S970).

FIG. 10 is a diagram illustrating a configuration of an embodiment of an apparatus for encoding / decoding a variable frame size having a constant transmission interval. Referring to FIG.

Referring to FIG. 10, the speech encoding apparatus 1000 according to the present invention operates as a transmitter and includes a variable encoder 1005 and a frame transmitter 1010. The voice decoding apparatus 1050 operates as a receiver and includes a frame receiver 1055 and a variable decoder 106.

The variable encoder 1005 determines an operation frame size according to the characteristics of the input voice, and encodes the input voice with the determined frame size.

Determination of the frame size according to the characteristics of the input voice has been described with reference to FIG. 1.

The variable encoder 1005 encodes the speech signal in a different frame size according to the characteristics of the input speech. The frame transmitter 1010 transmits speech data encoded in various sizes output by the variable encoder 1005 for each frame interval or at a constant transmission interval. The frame for this is shown in Fig. 11C.

The speech decoding apparatus 1050 performs a process opposite to that of the speech encoding apparatus 1000. That is, the frame receiving unit 1055 receives a frame transmitted at an irregular interval or a frame transmitted at a constant interval, and the variable decoding unit 1060 performs decoding according to the size of the received frame.

The concept of an encoding / decoding apparatus according to the present invention shown in FIG. 10A may be applied to each of the inventions shown in FIGS. 1, 6, and 8.

FIG. 10B is a flowchart illustrating a flow of an embodiment of a voice encoding method of a variable frame size having a constant transmission interval.

Referring to FIG. 10B, the variable encoder 1005 determines a frame size according to characteristics of an input voice, delay information of a network, and information about a type of a counterpart encoder, and encodes an input voice based on the determined frame size ( S1080).

The frame transmitter 1010 transmits a frame encoded in various sizes by the variable encoder 1005 at regular transmission intervals (S1090).

11 illustrates various frame types according to the present invention.

11A and 11B illustrate a structure of encoding and transmitting an input voice at regular intervals. For example, FIG. 11A has a frame size of 10 ms. In other words, the voice encoder is a reference speech encoder that always encodes the input speech signal in units of 10ms and transmits the unit in units of 10ms. FIG. 11B shows a conventional speech encoder having a frame size of 20 ms, encoding an input speech signal every 20 ms, and transmitting the apparatus in 20 ms units.

FIG. 11C is a diagram illustrating the features of the present invention shown in FIGS. 10A and 10B. The transmission interval is shown by a solid line, and the coding frame size is shown by a dotted line. Referring to FIG. 11C, the frame size of the speech encoder is encoded in 10 ms or 20 ms units according to the characteristics of the input speech signal, but transmission is always performed in 20 ms units. That is, the frame size for analyzing the input voice signal is determined according to the characteristics of the input voice signal, but the transmission may be at regular intervals.

FIG. 11D is a diagram illustrating the features of the present invention shown in FIGS. 1 to 9B, wherein a voice signal is encoded in units of 10 ms or 20 ms according to characteristics of an input voice, and a transmission interval is variably changed according to a variable frame size. The frame is shown. That is, the frame size of the speech encoder is determined according to the characteristics of the input speech signal, and the coded data is also transmitted at the predetermined frame size intervals.

The invention can also be embodied as computer readable code on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, and the like, and may also be implemented in the form of a carrier wave (for example, transmission over the Internet). Include. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

So far I looked at the center of the preferred embodiment for the present invention. Those skilled in the art will appreciate that the present invention can be implemented in a modified form without departing from the essential features of the present invention. Therefore, the disclosed embodiments should be considered in descriptive sense only and not for purposes of limitation. The scope of the present invention is shown in the claims rather than the foregoing description, and all differences within the scope will be construed as being included in the present invention.

According to the present invention, since the frame size, the quantizer structure and the bit allocation can be optimally adjusted according to the input speech, the performance of the speech encoding apparatus can be improved.

In addition, by adjusting the frame size of the voice coder according to the network state or the type of the voice coder of the other party, the quality of voice service can be improved by adjusting the delay required for voice data transmission.

Claims (20)

An input speech rating determiner classifying the rating of the input speech into a transition section and a static section; A variable adjusted speech encoder that variably performs encoding using a frame size, a quantizer structure, and bit-allocation corresponding to the determined class; And And a multiplexing unit for outputting a bit string of the input speech encoded with the variable frame size. The method of claim 1, And the voice grade determiner determines a grade of the input voice using an open-loop grade method or a closed-loop grade method. The method of claim 1, The variable adjustment speech encoder, An input voice temporary storage unit configured to store input voice samples by the size of a frame corresponding to the determined class; And And a variable speech encoder provided for each class of the input speech and encoding the input speech sample using a frame size, a quantizer structure, and bit-allocation corresponding to the determined grade. . (a) classifying an input speech into a transition section and a static section; (b) variably performing encoding using frame size, quantizer structure, and bit-allocation corresponding to the determined class; And and (c) outputting a bit string of the input speech encoded with the variable frame size. When receiving a bit stream encoded using a different frame size, quantizer structure, and bit-allocation according to the grade of an input speech classified into a transition period and a static period, the inverse of extracting parameter information necessary for decoding from the bit string Multiplexer; A variable-adjusted speech decoder provided for each grade of the input speech to variably perform decoding corresponding to the received grade of the input speech; And And a temporary storage unit for temporarily storing the decoded input speech to be continuously output. (a) When receiving a bit stream encoded using a different frame size, quantizer structure, and bit-allocation according to the grade of an input speech classified into a transition period and a static period, parameter information necessary for decoding is obtained from the bit string. Extracting; (b) variably performing decoding corresponding to the grade of the received input speech, provided for each grade of the input speech; And and (c) temporarily storing the decoded input speech to be continuously output. A frame determiner configured to determine a frame size having the same size as the frame size of the call counterpart encoder and the number of frames per packet based on information on the type of the call counterpart encoder; A variable adjusted speech encoder for variably encoding an input speech corresponding to the determined frame size and the number of frames; And And a multiplexer for outputting a bit string of the input speech encoded with the variable frame size. The method of claim 7, wherein And the frame determiner reduces the frame size and the number of frames when the network delay increases, and increases the frame size and the number of frames when the network delay decreases. delete The method of claim 7, wherein And the frame determiner determines the frame size and the number of frames based on delay information of a network that changes during a call. The method of claim 7, wherein And the frame determiner determines the frame size and the number of frames based on information on the speech encoder obtained when the call is connected to the call counterpart. The method of claim 7, wherein The variable adjustment speech encoder, An input voice temporary storage unit which stores an input voice sample by the size of the determined frame; And And a variable speech encoder configured to encode the input speech sample by using the speech encoder corresponding to the determined frame size among the speech encoders provided for each frame size. (a) determining a frame size and the number of frames per packet having the same size as the frame size of the encoder of the call counterpart, based on the information on the type of the encoder of the call counterpart; (b) variably encoding a speech signal corresponding to the determined frame size and the number of frames; And and (c) outputting a bit string of the speech signal encoded in the variable frame size. A demultiplexer configured to extract parameter information necessary for decoding from the bit string when receiving a bit string of a speech signal encoded according to a frame size and a frame number per packet determined based on delay information of a network; A variable-adjusted speech decoder provided for each frame size to variably perform decoding corresponding to the size of the received speech signal; And And a temporary storage unit for temporarily storing the decoded voice signal to be continuously output. (a) extracting parameter information necessary for decoding from the bit string when receiving the bit string of the speech signal encoded according to the frame size and the number of frames per packet determined based on delay information of the network; (b) variably performing decoding corresponding to the size of the received voice signal provided for each frame size; And and (c) temporarily storing the decoded speech signal to be continuously output. A variable encoder configured to determine a frame size for encoding based on any one of characteristics of an input voice, delay information of a network, and information about a type of an encoder of a call counterpart, and to encode the input voice based on the determined frame size; And And a frame transmitter configured to transmit the encoded frame at a constant transmission interval. The method of claim 16, The variable encoding unit divides the input speech into a transition section and a static section, and performs a speech encoding optimized according to the speech characteristics of each section. The method of claim 16, The variable encoding unit reduces the frame size when the delay of the network increases, and increases the size of the frame when the delay of the network decreases. The method of claim 16, And the variable encoding unit encodes the input speech in the same frame size as that of the encoder of the other party. Determining a frame size for encoding based on any one of characteristics of an input voice, delay information of a network, and information on a type of an encoder of a call counterpart, and encoding the input voice based on the determined frame size; And And transmitting the encoded frame at a constant transmission interval.

Images