JPH04923A - Variable rate codec - Google Patents

Abstract

PURPOSE:To improve the transmission efficiency of sound data by providing the encoding means for the sound data of the unit of a frame to be inputted, a sound detection means, a means which stores only the sound data, an encoding rate information generation means, a compression means, and an ATM cell preparing means. CONSTITUTION:An A/D circuit 211 inputs quantization data corresponding to each frame to an encoder part 221 and a sound detection part 241 respectively. The sound detection part 241 detects whether the corresponding frame is a sound part or a silent part. A buffer part 231 stores only the sound part as a burst data. A decoder 271 compresses encoding rate information, and a multiplexing part 281 multiplexes an adaptation header to be outputted from a header generation part 261 and the burst data to be outputted from the buffer part 231 and prepares an ATM cell.

Description

[Detailed Description of the Invention] [Summary] Regarding a variable rate codec that creates ATM cells by encoding voice data at a variable encoding rate, the purpose of this is to improve the transmission efficiency of voice data. An encoding means that performs encoding by setting an encoding rate for each frame of audio data in units of frames and outputs the encoded data; voice detection means for detecting whether or not there is a voice;
A storage means for storing only encoded data corresponding to a voiced part, and an encoding rate for generating and outputting encoding rate information expressing the encoding rate of each frame set by the encoding means in a predetermined number of bits. information generation means; compression means for compressing encoding rate information corresponding to a plurality of frames output from the encoding rate information generation means; and encoding rate information compressed by the compression means and encoded data stored in the storage means. and ATM cell creation means for creating and outputting an ATM cell containing the following information.

[Industrial application field]

The present invention relates to a variable rate codec that creates ATM cells by encoding voice data at a variable encoding rate.

[Conventional technology]

In recent years, ATM (Asynchronous Transfer), which has both the bandwidth flexibility of packet switching and the time transparency of circuit switching, has been used as a method for realizing high-bandwidth I5DN.
A transmission method using rMode+ (asynchronous transfer mode) has been agreed upon by the CCITT, and research is being actively conducted at various institutions.

A variable rate codec that uses such an ATM method quantizes an input audio signal, then encodes it at a variable rate, and then converts the result into one or more ATM cells and outputs it. be. For example, an ATM cell is created by using quantized data for every four frames as one frame, and using l multiframes made up of these six frames as a processing unit.

FIG. 4 shows an outline of the operation of the variable rate codec.

The variable rate codec detects only the voiced portion of the input voice and encodes only the quantized data of the frame corresponding to the voiced portion at a variable rate to obtain burst data corresponding to one multiframe. Variable encoding rates include, for example, 16, 24, 32, 40, and 48K.
Assuming that five types of bps are considered, the data amount of the burst data described above differs depending on which encoding rate among these is used for encoding and which frame is a sound portion. Therefore, if the amount of burst data is small, 1
If the amount of data is large, it is converted into two or three ATM cells.

By the way, by performing variable rate encoding in this way, AT
When an M cell is created, it is necessary to include information regarding the encoding rate corresponding to each frame and which frame corresponds to a voiced part in the ATM cell. FIG. 5 shows the format of the ATM cell, and FIG. 6 shows the details of the coding rate information and speech/silence information contained in the ATM cell.

As shown in FIG. 5, an ATM cell consists of a 5-byte ATM header containing control information necessary for ATM exchange, such as destination, and a 48-byte fixed-length data area for storing burst data.

In addition, the first 6 bytes of the data area are allocated to a control information area called an adversion header that is sent and received between variable rate codecs. 1 byte, 3 bytes for encoding rate information, 1 byte for voice/silence information, and 1 hide for frame head position information. The frame head position information is information indicating the frame head position within the data area.

Furthermore, as shown in Figure 6, the encoding rate information indicates at which encoding rate the burst data was processed, taking into account the case where all frames are sound parts. , 3 bits are allocated to each frame, for a total of 18 bits.

In the example shown in the figure, three frames correspond to the sound portion, so the encoding rate of each of these frames is expressed by the lower 9 bits of the encoding rate information (3 bits x 3 frames). There is. Furthermore, the sound/silence information indicates whether each frame constituting one multiframe is a sound part or a silent part, and "1" is set for a sound part, and °0 is set for a silent part. Stored.

[Problem to be solved by the invention]

By the way, in the conventional method described above, 18 bits, or 3 bytes, are used for the data area to represent five types of encoding rates for one multiframe, and the capacity of the data area for storing burst data is correspondingly reduced. There was a problem that the transmission efficiency of audio data deteriorated because the number of audio data decreased.

In the ATM cell shown in Figure 5, 6 bytes of the 48-byte data area are used as an adaptation header, so coded data is stored in the remaining 42-byte data area. .

The present invention was created in view of these points, and an object of the present invention is to provide a variable rate codec that can improve the transmission efficiency of audio data.

[Means to solve the problem]

FIG. 1 is a principle block diagram of the variable rate codec of the present invention.

In the figure, encoding means 111 encodes input audio data in units of frames by setting a encoding rate for each frame, and outputs encoded data.

The audio detection means 121 detects whether or not there is a sound for each frame of the input audio data.

The storage means 131 stores only encoded data corresponding to the voiced portion.

The encoding rate information generating means 141 is the encoding means 111
It generates and outputs encoding rate information that expresses the encoding rate of each frame set in a predetermined number of bits.

Compressing means 151 compresses encoding rate information corresponding to a plurality of frames output from encoding rate information generating means 141.

The ATM cell creation means 161 creates and outputs an ATM cell containing the encoding rate information compressed by the compression means 151 and the encoded data stored in the storage means 121.

Therefore, the overall arrangement is to compress the coding rate information within the ATM cell.

[For production]

When audio data is input, it is encoded by the encoding means 111 at a different encoding rate for each frame, and only the encoded data of the sound portion detected by the voice detection means 121 is stored in the storage means 131. . Further, when the encoding rate information for notifying the audio data transmission destination of the encoding rate for each frame described above is generated by the encoding rate information generating means 141, it is further compressed by the compression means 151, and this compressed An ATM cell containing coding rate information is created.

In the present invention, the encoding rate information is compressed and
By creating cells, it is possible to include more coded data in a fixed length ATM cell, thereby increasing the transmission efficiency of voice data.

[Example] Hereinafter, an example of the present invention will be described in detail based on the drawings.

FIG. 2 shows the configuration of a variable rate codec in one embodiment of the present invention. In the figure, 211 is an analog-digital conversion circuit (A/D circuit), 221 is an encoder section, 231 is a buffer section, 241 is an audio detection section,
251 is an encoding control unit, 261 is a header generation unit, 2
Reference numeral 81 indicates a multiplexing section.

The A/D circuit 211 quantizes the input analog audio signal at a predetermined sampling frequency.
Quantized data corresponding to each frame of s period is input to the encoder section 221 and the speech detection section 241, respectively.

The encoder section 221 encodes the input quantized data of each frame at a variable rate, and the encoding rate is 16, 24, 32, 40, 48.
Five types of Kbps are available. As the encoding method, for example, an adaptive differential coding method (ADPCM) is used. The audio detection unit 241 detects whether the corresponding frame is a sound portion or a silent portion of the audio signal, based on the input quantized data.

The buffer unit 231 stores only the encoded data processed by the encoder unit 221 that corresponds to the sound portion as burst data.

This sound portion is designated by the encoding control unit 251.

When the audio detection unit 241 detects a voiced part in the encoded data, the encoding control unit 251 sends an instruction to store the encoded data to the buffer unit 231, and also sends an instruction to store the encoded data to the buffer unit 231. Based on information such as whether there is sound or silence, and if there is sound, the encoder unit 221 performs encoding at which encoding rate, etc.
The header generation unit 261 is instructed to generate an adaptation header.

The header generation unit 261 is for creating various information in the adaptation header. Header generator 261
263 is a cell number generation section, 265 is a voice/silence information generation section, 267 is a frame head position information generation section, 269 is a coding rate information generation section, 271 is a decoder, 273 indicate selectors.

The cell number generation unit 263 generates a cell number that is a serial number assigned to each cell when burst data corresponding to one multiframe is divided into a plurality of ATM cells and stored. The cell number is output as 1 hide data.

The voice/silence information generation unit 265 generates voice/silence information indicating whether each frame constituting one multiframe corresponds to a voice portion or a silent portion of audio. This sound/non-sound information is output as 1-byte data containing 6-bit data in which one frame corresponds to one bit.

The frame start position information generation unit 267 generates frame start position information indicating the start position of each frame within the data area. The frame head position information is output as 1-byte data.

The encoding rate information generation unit 269 generates encoding rate information representing the encoding rate corresponding to each frame constituting one multiframe; 3 pinpoints corresponding to each frame to represent 5 types of encoding rates. The data, ie, a total of 18 pinto data, is output as encoding rate information.

The decoder 271 compresses (encodes) this 18-bit encoding rate information into 8 bits.

For example, when encoding at 16 Kbps, 1 frame (1
(assuming a multi-frame period of 241 m5) is converted to 8-byte burst data, and when encoded at 48 Kbps, becomes 24-byte burst data.

Therefore, the number of frames that can fit in one ATM cell is 5 even when the encoding rate of all frames is 16 Kbps, and the number of frames decreases as the encoding rate becomes thicker. If we focus only on the sound part of the combinations of different types of encoding rates, there are a total of 239 combinations of encoding rates corresponding to one ATM cell, which can be compressed into 1 byte (2' = 256). Become.

The selector 273 selects each 1-byte data output from each of the generation units 263 to 267 and the decoder 271 described above in the order of cell number, encoding rate information, voice/silence information, and frame head position information, and converts the data into 4 bytes. output as an adaptation header.

The multiplexer 281 multiplexes the adaptation header output from the ladder generator 261 and the burst data output from the buffer 231 to create an ATM cell. The ATM cell created by the multiplexer 281 has the format shown in FIG. 5, in which the data area excluding the adaptation header has a capacity of 44 bytes.

Note that the ATM header is added to the data area by the multiplexer 281 to create an ATM cell.

FIG. 3 shows an outline of processing when encoding rate information is compressed using the variable rate codec described above.

The voice detection unit 241 detects a voice part of the voice signal, and burst data is created using only the variable rate encoded data corresponding to this voice frame, and the burst data is sent to the buffer unit 23.
It is stored in 1. The breakdown of this burst data is, for example, encoded data of 16 Kbps, 32 Kbps, and 24 Kbps as shown in the figure. The encoding rate information generation unit 269 in the header generation unit 261 generates these three
It generates and outputs 18-bit encoding rate information including a total of 9 bits representing each of the 3 encoding rates in its lower order. The reason for including them in the lower order is to reduce the number of combinations of coding rates by removing silent frames sandwiched between sound frames.

Further, even if the encoding rate of the voice frame is reduced in this way, there is no particular problem because it can be made to correspond to an accurate frame position based on the voice/silence information.

The decoder 271 compresses this 18-bit encoding rate information into 1 byte, and an ATM cell containing this compressed encoding rate information is output from the variable rate codec.

Furthermore, since the 1-byte encoding rate information compressed in this way has a one-to-one correspondence with the 18-bit encoding rate information before compression, the variable rate codec (not shown) that received the ATM cell ), on the other hand, performs decompression (decoding processing) to reproduce the 18-bit encoding rate information,
After performing variable rate decoding processing, dequantization is performed to obtain an analog audio signal.

In this way, by compressing the 18-bit encoding rate information for 6 frames generated by the encoding rate information generation unit 269 to 8 bits by the decoder 271, the capacity that conventionally required 3 bytes is reduced. The encoding rate information is 1
Expressed in bytes, the effective part of the data area (the data area excluding the adversion header) can be increased by the amount of the reduction, making it possible to increase the transmission efficiency of audio data.

In the embodiment of the present invention described above, a case was considered in which five types of encoding rates and one multiframe were composed of six frames, but the present invention can be applied to other combinations as well. Even in this case, it is sufficient to analyze the combination patterns of encoding rates that actually exist, and compress the encoding rate information to the minimum number of bits that can represent this number of beats.

〔Effect of the invention〕

As described above, according to the present invention, by compressing the encoding rate information to create an ATM cell, more encoded data can be included in a fixed-length ATM cell, thereby increasing the transmission efficiency of voice data. It is extremely useful from a practical point of view.

[Brief explanation of drawings]

FIG. 1 is a principle block diagram of a variable rate codec of the present invention, FIG. 2 is a block diagram of a variable rate codec according to an embodiment of the present invention, and FIG. 3 is an explanatory diagram of encoding rate information compression in an embodiment. FIG. 4 is a diagram showing an outline of the operation of the variable rate codec, FIG. 5 is an explanatory diagram of ATM self-formatting, and FIG. 6 is an explanatory diagram of conventional encoding rate information. In the figure, 111 is an encoding means, 121 is a voice detection means, 131 is a storage means, 141 is an encoding rate information generation means, 151 is a compression means, 161 is an ATM cell creation means, and 211 is an analog-digital conversion circuit (A /D circuit)
, 221 is an encoder section, 231 is a buffer section, 241 is a voice detection section, 251 is an encoding control section, 261 is a header generation section, 263 is a cell number generation section, 265 is a voice/silence information generation section, 267 269 is a frame head position information generating section, 269 is a coding rate information generating section, 271 is a decoder, 273 is a selector, and 281 is a multiplexing section. 1@(!l L-N Multi-frame~p set child 1tenge b7n number Ishib-1t-'s departure from the sun Fig. 3 Fig.

Claims (1)

[Claims] (1) An encoding means (111) that sets and encodes an encoding rate for each frame of input audio data in units of frames, and outputs the encoded data; On the other hand, a voice detection means (121) that detects whether or not there is a sound for each frame, a storage means (131) that stores only the encoded data corresponding to the sound part, and an encoding means ( Encoding rate information generation means (141) that generates and outputs encoding rate information expressing the encoding rate of each frame set in step 111) in a predetermined number of bits.
and a compression means (151) for compressing the encoding rate information corresponding to a plurality of frames outputted from the encoding rate information generating means (141), and the encoding rate information compressed by the compression means (151) and the aforementioned encoding rate information. A variable rate codec comprising: ATM cell creation means (161) for creating and outputting an ATM cell containing encoded data stored in a storage means (131).