JPH10340099A - Audio decoder device and signal processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an audio decoder device capable of efficiently using a memory bus. SOLUTION: A down mixing arithmetic unit 106 generates PCM data from the audio decoding data in each channel in an external storage 100, interleaves the respective PCM data, decreases a number of channels (down mixing) and stress PCM data of decreased channels in an external memory 100. After storing all auditory decoding data of the respective channels in the external memory 100 by performing decoding processing of a first block, in the course of decoding processing of a second block, auditory decoding data of each channel of the first block are down-mixed by dividing them into plural times.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio decoder for decoding PCM data from a coded bit stream in an AV (audio / visual) device, and further to a signal processing for such an audio decoder. It concerns the device.

[0002]

2. Description of the Related Art A conventional audio decoder device will be described with reference to FIGS.

FIG. 6 is a block diagram showing a configuration of a conventional audio decoder device. In FIG. 6, 500 is an external storage device, 501 is a bitstream parser,
502 is an exponent decoder, 503 is a mantissa data bit allocator, 504 is a mantissa decoder, and 505 is an IMDCT
506 is a down-mix operation unit, 507 is an internal storage device, and 508 is an integrated semiconductor device.

The bit stream is stored in an external storage device 500
, And then input to the bitstream parser 501. The bitstream parser 501 parses the bitstream, extracts data necessary for decoding, and outputs the data to the exponent decoder 502. The exponent part decoder 502 generates exponent data in the frequency domain from data necessary for the decoding process, and outputs the exponent data to the mantissa data bit allocator 503 and the IMDCT unit 505. The mantissa data bit allocator 503 generates exponent data in the frequency domain,
The mantissa bit allocation amount is calculated from the data stored in the external storage device 500, and the allocation amount is calculated by the mantissa decoder 50.
4 is output. The mantissa decoder 504 generates mantissa data in the frequency domain from the assigned amount,
Output to 05. The IMDCT 505 generates audio decoded data in the time domain from the exponent data and the mantissa data in the frequency domain, and stores the audio decoded data in the external storage device 500. The downmix calculator 506 generates PCM data from the audio decoded data stored in the external storage device 500, performs interleaving, and performs
To be stored. This PCM data is stored in the external storage device 500
Output from

FIG. 7 is a memory mapping of an external storage device of a conventional audio decoder device. 7, reference numeral 600 denotes an area for storing one block of PCM data, 601 denotes an area for storing one block of audio decoded data of channel 0, 602 denotes an area for storing one block of audio decoded data of channel 1, and 603.
Is an area for storing one block of audio decoded data of channel 2, 604 is an area for storing one block of audio decoded data of channel 3, 605 is an area for storing one block of audio decoded data of channel 4, and 606 is a channel. 5 is an area for storing one block of audio decoded data.

FIG. 8 is a flowchart showing a control method for decoding audio code data of each channel of one block in a conventional audio decoder device.

First, initialization of the register, the internal storage device 507, and the external storage device 500 is performed (step S1).
1). Then, a code data receiving process for inputting the bit stream stored in the external storage device 500 is performed (step S12).

[0008] Thereafter, the bit stream is parsed and a bit stream parsing process for extracting data necessary for the decoding process is performed (step S13). The index data in the frequency domain is generated using the extracted data (step S14). The mantissa bit allocation amount is calculated using the generated frequency domain index data (step S15). Using this mantissa bit allocation amount, mantissa data in the frequency domain is generated (step S1).
6). An IMDCT (Inversed Modified Discrete Cosine) for generating audio decoded data using the generated mantissa data in the frequency domain and exponent data in the frequency domain.
Transform) (Step S17). The generated audio decoded data is stored in the external storage device 500 (step S18).

The above process is repeated for each channel included in one block (step S19).
Thus, audio decoded data corresponding to each channel included in one block is generated and can be stored in the external storage device 500.

[0010] Thereafter, the audio decoded data of each channel of one block in the external storage device 500 is input (step S20), and the down-converted data for converting the audio decoded data of each channel of one block into PCM data of one block is input. Perform mix operation processing (step S2
1). Then, the PCM data of one block is output from the external storage device 500 (step S22).

As described above, in the conventional audio decoder device, since one block of PCM data is calculated by one downmix operation, audio decoding data is input from the external storage device 500 as a pre-process, The amount of data transferred during the process of writing PCM data to the external storage device 500 becomes very large, occupying the memory bus, and the external storage device 500
There is a problem that it interferes with other processes that use.

Next, a part of the code data of a plurality of channels may be shared by the respective channels.
For example, high-frequency band code data common to at least one channel included in at least one of the channels is decoded to form high-frequency band decoded data, and low-frequency code data is decoded for each channel. To form decoded data in the low frequency band, and combine the decoded data in the low frequency band with the decoded data in the high frequency band, thereby forming decoded data for each channel.

Such a decoding process will be specifically described with reference to FIGS.

In this decoding process, a multi-channel audio signal is expressed in the frequency domain, and an encoded bit stream including a mantissa part and an exponent part is input. The encoded bit stream is included in at least one of a plurality of channels. The high frequency band code data common to each channel is decoded to form high frequency band decoded data, and the low frequency band code data is decoded for each channel to form low frequency band decoded data. Band decoded data is combined with the high frequency band decoded data, thereby forming decoded data for each of the channels.

FIG. 20 shows a conventional signal processing device for performing the above-mentioned decoding processing. In FIG. 20, a bit stream is temporarily stored in an internal storage device 1301 of the signal processing device, and a bit stream parser 130
0 is analyzed to extract necessary data. Based on the extracted data, exponent data is generated by the exponent part decoder 1302 in the frequency domain. Based on the exponent data, the mantissa data bit allocator 1303 determines the bit allocation amount to the mantissa data. The mantissa data is generated by the mantissa decoder 1304 in the frequency domain based on the bit allocation amount to the mantissa data.
Based on the data generated from the exponent part decoder 1302 and the mantissa part decoder 1304, a frequency domain data generator 1305
To generate frequency domain data.

In the frequency domain data generator 1305, when decoding code data of an arbitrary channel,
First, high-frequency code data common to a plurality of channels included in a predetermined channel is decoded to obtain high-frequency band decoded data.
The ratio between the signal power of a predetermined channel and the signal power of an arbitrary channel obtained by the encoder is multiplied, and the result is combined with the decoded data of the low frequency band of the arbitrary channel.
Thereby, decoded data of an arbitrary channel is obtained.

The decoded data in the frequency domain is converted into decoded data in the time domain by the time domain converter 1306,
The result is converted to PCM data and output.

FIG. 21 shows an outline of the decoding process of the code data of an arbitrary channel.

When a predetermined channel 1400 is decoded, a decoded data band 1402 of a low frequency band of a low frequency band is obtained.
And decoded data 1 in a high frequency band common to a plurality of channels.
403 is formed (step 141). The decoded data 1403 in the high frequency band of the arbitrary channel 1401 is multiplied by the ratio α of the signal power of the predetermined channel 1400 and the signal power of the arbitrary channel 1401 to the decoded data 1403 in the high frequency band. 1404 is generated (step 142). The decoded data 1405 of the low frequency band of the arbitrary channel 1401 is combined with the decoded data 1404 of the high frequency band to generate the decoded data of the channel 1401 (step 143).

As described above, if high-frequency code data common to a plurality of channels is used, it is not necessary to transmit high-frequency code data for each channel, so that transmission efficiency can be improved.

In such decoding processing, necessary data is extracted from the bit stream stored in the internal storage device 1301 while indicating the bit stream by a plurality of pointers. This will be described next with reference to FIG.

After decoding a predetermined channel 1400, the mantissa part 1202 and the exponent part 1201 of the low-frequency code data of an arbitrary channel 1401 existing in the bit stream 1200 are read out while being indicated by each pointer, and the low-frequency code data is read out. The code data is decoded, and the mantissa part 1202 and the exponent part 1201 of the high-frequency code data of the predetermined channel 1400 are read out while being indicated by the pointers 1203 and 1204, and the high-frequency code data is decoded.

For this reason, each of the pointers 1203 and 1204
Has to be controlled so as to rewind as indicated by arrows 1205 and 1206. Further, it is necessary to hold the bit stream in the storage device until the decoding process of all the channels sharing the high-frequency code data is completed. Therefore, in order to perform the decoding process of each channel sharing the high-frequency code data, it is premised that a large storage capacity for holding a bit stream is provided.

Further, since the decoding process of the high-frequency code data has a larger load than the decoding process of the normal low-frequency code data,
Treatment mitigation measures were required.

[0025]

As described above, in the conventional audio decoder device, one block of P
Since the CM data is calculated by one downmix operation, the amount of data transferred to the external storage device 500 becomes very large, occupying the memory bus, and hindering other processes using the external storage device 500. There was a problem of giving.

In the conventional signal processing device, it is necessary to control the pointers 1203 and 1204 so as to rewind. Further, it is necessary to hold the bit stream in the storage device until the decoding process of all the channels sharing the high-frequency code data is completed. Therefore, in order to perform the decoding process of each channel sharing the high-frequency code data, it is premised that a large storage capacity for holding a bit stream is provided. Further, the decoding process of the high-frequency code data has a larger load than the decoding process of the normal low-frequency code data.

Therefore, a first object of the present invention is to provide an audio decoder device that can use a memory bus efficiently.

A second object of the present invention is to reduce decoding processing of code data common to each channel, and to hold code data of all channels in a storage device until the decoding processing is completed. An object of the present invention is to provide an unnecessary signal processing device.

[0029]

In order to solve the above-mentioned problems, an audio decoder apparatus of the present invention inputs a bit stream in block units, decodes one block of bit stream, and performs audio decoding of a plurality of channels. An audio decoder device for forming data and downmixing the audio decoded data of each channel, comprising a storage unit for storing audio decoded data of each channel corresponding to a bit stream of a first block, wherein a bit stream of a second block is provided. In the decoding process, the audio decoding data of each channel corresponding to the bit stream of the first block in the storage unit is downmixed.

In one embodiment, the bit stream of the second block is converted into audio decoded data of each channel by a plurality of decoding processes. Each time the decoding process is performed, the first block in the storage means is converted. , The audio decoded data of each channel corresponding to the bit stream is divided and sequentially downmixed.

In one embodiment, the bit stream of the second block is converted into audio decoded data for each of the channels by repeating the decoding process for each of the channels, and the decoding process of each of the channels is performed. Each time, the audio decoding data of each channel corresponding to the bit stream of the first block in the storage unit is divided and sequentially downmixed.

In one embodiment, the audio decoded data formed by the downmixing is temporarily stored in the storage means and then output.

Further, the audio decoder device of the present invention converts a multi-channel audio signal into each frequency domain, and then inputs and decodes a bit stream formed by encoding a mantissa and an exponent. A bit stream parser for parsing the bit stream and extracting data required for decoding processing, an internal storage device for storing data required for decoding processing, and an internal storage device. An exponent part decoder that generates exponent data in the frequency domain of the audio signal based on the data of the audio signal, a mantissa data bit allocator that calculates a mantissa bit allocation amount from the exponent data output from the exponent part decoder, Based on the allocation amount output from the mantissa data bit allocator, the audio A mantissa decoder for generating mantissa data in the frequency domain of the signal, and the exponent decoder and the exponent data and mantissa data generated by the mantissa decoder are converted from the frequency domain to the time domain. Accordingly, an IMDCT unit that forms the audio decoded data of each channel, a downmix calculator that generates and interleaves PCM data from the audio decoded data of each channel, and a bit stream that is generated by the IMDCT unit An external storage device for storing the window delay data, the audio decoded data, and the PCM data, wherein the bit stream is input in block units, and when the bit stream of the second block is decoded, the external storage device is The stored bit of the first block PCM from decoded audio data for each channel corresponding to the stream
Generate data.

In one embodiment, the external storage device is P
The PCM data storage area includes a CM data storage area and an audio decoded data storage area corresponding to each channel. The PCM data storage area has a capacity capable of storing PCM data corresponding to one block including a data amount of a plurality of channels × a plurality of data. The audio decoded data storage area is divided into areas corresponding to respective channels, and each area has a capacity capable of storing the audio decoded data exceeding the one block of the corresponding channel.

In one embodiment, an audio decoded data write pointer corresponding to each channel for writing the audio decoded data to the external storage device, and an audio decoded data write pointer corresponding to each channel for reading the audio decoded data from the external storage device. An audio decoded data read pointer, a PCM write pointer for writing the PCM data to the external storage device, an audio decoded write pointer, and an audio decoded data storage area corresponding to each channel for updating the audio read pointer. Address data and audio decoded data pointer return data, wherein the audio decoded data write pointer and the audio decoded data read pointer are independently updated and assigned to each channel. To cycle through.

In one embodiment, the downmix calculator divides the audio decoded data of each channel into N times and executes the process.

Next, the signal processing apparatus of the present invention inputs a bit stream including code data of a plurality of channels,
The common code data common to each channel included in at least one of the channels is decoded to form common decoded data, and the channel code data unique to the channel is decoded for each channel. A signal processing device for forming decoded data, combining the channel decoded data with the common decoded data, thereby forming decoded data of each channel, and decoding the common code data to form common decoded data. Storage means for storing the common decoded data from the storage means each time the channel code data unique to the channel is decoded to form channel decoded data, and the common decoded data and the channel decoded data are read from the storage means. And control means for performing the connection.

Further, the signal processing device of the present invention inputs a bit stream including code data of a plurality of channels,
The common code data common to each channel included in at least one of the channels is decoded to form common decoded data, and the channel code data unique to the channel is decoded for each channel. A signal processing device for forming decoded data, combining the channel decoded data with the common decoded data, thereby forming decoded data of each channel, and storing intermediate data in decoding the common code data in the signal processing device. And, each time the channel code data unique to the channel is decoded to form channel decoded data, the intermediate data is read from the storage means,
A control unit is provided for forming the common decoded data from the intermediate data and for combining the common decoded data with the channel decoded data.

Further, the signal processing apparatus of the present invention converts a multi-channel audio signal into each frequency domain, and then inputs a bit stream formed by encoding to represent a mantissa part and an exponent part, Decoding high-frequency band code data common to each channel included in at least one of the above to form high-frequency band decoded data, and decoding low-frequency band code data for each of the channels to perform low-frequency band decoding Forming a data, combining the low frequency band decoded data with the high frequency band decoded data, thereby forming a decoded data of each channel in a signal processing device, an external storage device for temporarily storing the bit stream; Input stream parsing for parsing the bit stream and extracting data required for decoding processing An internal storage device for storing data necessary for decoding processing; an exponent part decoder for generating frequency domain exponent data of the audio signal based on the data in the internal storage device; A mantissa data bit allocator for calculating a mantissa bit allocation amount from exponent data output from the device, and generating mantissa data in a frequency domain of the audio signal based on the allocation amount output from the mantissa data bit allocation device. From the mantissa part decoder, the exponent part decoder and the exponent data and the mantissa data generated by the mantissa part decoder, synthesize the high frequency band decoded data and the low frequency band decoded data of each channel, and The low frequency band decoded data is combined with the high frequency band decoded data to perform conversion from the frequency domain to the time domain. A data generator for forming the decoded data of each channel, wherein the high-frequency band decoded data is stored in the internal storage device, and when forming the low-frequency band code data of the channel, the internal storage is used. Reading the high frequency band decoded data from the device and combining the low frequency band code data with the high frequency band decoded data.

In one embodiment, the high-frequency band decoded data is compressed and stored in the internal storage device.

Further, the signal processing apparatus of the present invention converts a multi-channel audio signal into a frequency domain and then inputs a bit stream formed by encoding to represent a mantissa part and an exponent part, Decoding high-frequency band code data common to each channel included in at least one of the above to form high-frequency band decoded data, and decoding low-frequency band code data for each of the channels to perform low-frequency band decoding Forming a data, combining the low frequency band decoded data with the high frequency band decoded data, thereby forming a decoded data of each channel in a signal processing device, an external storage device for temporarily storing the bit stream; Input stream parsing for parsing the bit stream and extracting data required for decoding processing An internal storage device for storing data necessary for decoding processing; an exponent part decoder for generating frequency domain exponent data of the audio signal based on the data in the internal storage device; A mantissa data bit allocator that calculates a mantissa bit allocation amount from exponent data output from the device, and, based on the mantissa bit allocation amount output from the mantissa data bit allocation device, calculates From the mantissa decoder to generate, from the exponent data and the mantissa data generated by the exponent decoder and the mantissa decoder, synthesize the high frequency band decoded data and the low frequency band decoded data of each channel, The low-frequency band decoded data of each channel is combined with the high-frequency band decoded data to transform from the frequency domain to the time domain. By performing, and a data generator for forming decoded data for the respective channels, stores the intermediate data in the internal storage device in the encoding processing of the high frequency band encoded data,
When forming the low-frequency band code data of the channel, the intermediate data is read from the internal storage device, the high-frequency band decoded data is formed from the intermediate data, and the low-frequency band code data is converted to the high-frequency band decoded data. To be combined.

In one embodiment, the intermediate data is compressed and stored in the internal storage device.

In one embodiment, the intermediate data is exponential data output from the exponent part decoder.

In one embodiment, the intermediate data is a mantissa bit allocation amount output from the mantissa data bit allocator.

In one embodiment, the intermediate data is mantissa data in the frequency domain output from the mantissa decoder.

[0046]

BEST MODE FOR CARRYING OUT THE INVENTION

(Embodiment 1) An embodiment of an audio decoder device of the present invention will be described with reference to FIGS. 1, 2, 3, 4, and 5. FIG.

FIG. 2 shows the configuration of a bit stream that is input to and decoded by the audio decoder device of the present embodiment. This bit stream is synchronized with the synchronization signal SYN.
C, error check signal CRC, system information signal S
I, stream information signal BSI, and audio blocks AB0, AB1, AB2, AB following each signal
3, AB4, AB5, etc.

Each audio block AB0, AB1, AB
2, AB3, AB4, and AB5 each include audio code data of up to six channels.

Among the channels, there is a channel called a normal channel. As audio code data, each exponent part Exp of up to 50 subbands (up to 253) is used.
And each mantissa Mant.

Each exponent part Exp and each mantissa part Mant for 50 subbands included in the audio code data of one channel are decoded to form exponent data and mantissa data in the frequency domain. Decoded data in the frequency domain is formed from the mantissa data, IMDCT is performed on the decoded data in the frequency domain, and conversion from the frequency domain to the time domain is performed to form decoded audio data in the time domain.

Among the channels, there is a channel called a basic channel, which is composed of a high frequency band and a low frequency band, and further includes combined data cpl. With this combined data cpl, each exponent part Exp for 50 subbands
Then, each mantissa Mant is divided into a low frequency band and a high frequency band. Each exponent part Exp and each mantissa part Mant corresponding to the high frequency band are extracted from the audio code data of the channel, and are also given to audio code data of some other channels.

Among the channels, there is a channel called a coupling channel, which is composed of only a low-frequency band by separating a high-frequency band in advance, and each exponent part Exp and each mantissa part Mant corresponding to this low-frequency band. including. The respective exponent parts Exp and the respective mantissa parts Mant are decoded to form low frequency band exponent data and mantissa data, and are converted from the frequency domain to the time domain to form audio decoded data in the time domain. I do. Further, based on each exponent part Exp and each mantissa part Mant corresponding to the high frequency band of the basic channel, audio decoded data of a high frequency band is formed. The low-frequency band and high-frequency band audio decoded data are combined to form one-channel audio decoded data.

Alternatively, among the channels, there is a channel referred to as a low-frequency channel, which does not originally have a high-frequency band but is composed of only a low-frequency band. Decoding the mantissa part Mant to form exponent data and mantissa data of the low frequency band,
The conversion from the frequency domain to the time domain is performed to form audio decoded data for one channel.

FIG. 1 is a block diagram showing the configuration of the embodiment of the present invention. In FIG. 1, 100 is an external storage device, 101 is a bit stream parser, 102 is an exponent decoder, 103 is a mantissa data bit allocator, 1
04 is a mantissa decoder, 105 is frequency data and IMD
A CT device, 106 is a downmix calculator, 107 is an internal storage device, and 108 is an integrated semiconductor device.

The bit stream is stored in the external storage device 100
, And then input to the bitstream parser 101. The bitstream parser 101 parses the bitstream, extracts each exponent part Exp of a predetermined channel from one block, and outputs each exponent part Exp to the exponent part decoder 102.

The exponent part decoder 102 generates exponent data in the frequency domain based on each exponent part Exp, and outputs the exponent data in the frequency domain to the mantissa data bit allocator 10.
3 and output to the frequency data and IMDCT device 105.

The mantissa data bit allocator 103 calculates the mantissa bit allocation amount based on the exponent data in the frequency domain and each mantissa part Mant of a predetermined channel of one block read from the bit stream in the external storage device 100. , And outputs the mantissa bit allocation to the mantissa decoder 104.

The mantissa decoder 104 generates mantissa data in the frequency domain from the mantissa bit allocation amount, and outputs it to the frequency data and the IMDCT device 105.

Frequency data and IMDCT device 105
Forms decoded data in the frequency domain from exponent data and mantissa data in the frequency domain, performs conversion from the frequency domain to the time domain, generates decoded audio data in the time domain, and stores the data in the external storage device 100.

The time domain audio decoded data is obtained for each channel, and the time domain audio decoded data is stored in the external storage device 100.

At this time, as described above, for the normal channel, each exponent part Exp and each mantissa part Mant for 50 subbands are decoded to form data in the frequency domain, and from the frequency domain to the time domain. To form audio decoded data.

Similarly, for the basic channel, each exponent part Exp and each mantissa part Mant are decoded to form data in the frequency domain, and conversion from the frequency domain to the time domain is performed to form decoded audio data. .

For the combined channel from which the high-frequency band has been separated in advance, each exponent part Exp and each mantissa part Mant corresponding to the low-frequency band included in the channel are decoded, and the high-frequency band included in the basic channel is decoded. Based on the data obtained by decoding the corresponding exponent parts Exp and the mantissa parts Mant, data of the entire frequency band is formed, and conversion from the frequency domain to the time domain is performed to form audio decoded data.

Further, for a low-frequency channel having no high-frequency band, each exponent part Exp and each mantissa part Mant corresponding to the low-frequency band included in the channel are decoded to form data of the low-frequency band. Then, conversion from the frequency domain to the time domain is performed to form audio decoded data.

The downmix calculator 106 generates PCM data from the audio decoded data of each channel in the external device storage device 100, performs interleaving on each PCM data, and reduces the number of each channel. (Downmix), the PCM data of the reduced channel is stored in the external storage device 100. Thereafter, the PCM data of the reduced channel is output from the external storage device 100.

In this embodiment, as described later in detail, the first
After the block decoding process is performed and the audio decoded data of all the channels is stored in the external storage device 100, the audio decoded data of each channel of the first block is divided into a plurality of times during the decoding process of the second block. Divide and mix down.

FIG. 3 shows the external storage device 100 of this embodiment.
It is a memory mapping in.

In FIG. 3, reference numeral 200 denotes one block of PC.
Area for storing M data, 201 is channel 1.
An area for storing 75 blocks of decoded audio data,
202 is an area for storing 1.75 blocks of audio decoded data of channel 1, 203 is an area for storing 2.75 blocks of audio decoded data of channel 2, and 204 is an area for storing 4.25 blocks of audio decoded data of channel 3. Storage area, 205 is channel 4
Area for storing four blocks of audio decoded data
Reference numeral 206 denotes an area for storing 1.5-block audio decoded data of channel 5. Each area does not need to be arranged in order.

The storage area corresponding to each channel is
In the middle of the decoding process of the block, it has a minimum capacity necessary to downmix the audio decoded data of each channel of the first block in the storage area in a plurality of times. Further, in each of the channels 3 to 5, the capacity of the storage area corresponding to each of the channels 3 to 5 is set to be large because the delayed audio decoded data is used.

The memory mapping shown in FIG. 3 is merely an example, and the size (storage capacity) of the area for storing the audio decoded data is appropriately set for each channel according to the amount of delay and other conditions. be able to. For example, when the delayed audio decoded data is not used in any of the channels, the size of the area of each channel is minimized, and each of the channels 0, 1, and 2. Are set in a 1.25 block area, channel 4 is set in a 1.00 block area, and channel 5 is set in a 1.5 block area.

FIG. 4 shows a method of accessing the external storage device 100. Here, as an example, it is assumed that the audio decoded data of each channel of one block is divided into four times and the downmix calculation process is performed.
The transition of the data read pointer for reading the audio decoded data of channel 0 is shown.

In the initial setting process, 0x1000h is set in the audio decoded data read pointer, and 0x17 is set in the audio decoded data storage area final address data.
00h, 0x700h is set in the audio decoded data pointer return data. In the process of inputting audio decoded data from the external storage device 100, which is a pre-process of the downmix operation process, the address of the external storage device 100 is determined with reference to the audio decoded data read pointer, and the read process is performed. Then, after the read processing is completed, the audio decoding data read pointer is updated for the next read processing.

In the updating method of the audio decoded data read pointer, first, the read data amount (0 × 100h) is added to the audio decoded data read pointer. next,
In order to determine whether the added audio decoded data read pointer is within the allocated storage area of the external storage device 100, the read pointer is compared with the audio decoded data storage area final address data (0x1700h). If it is within the range, the added value is used as it is. If it is out of the range (when the added audio decoded data read pointer is larger or equal), the audio decoded data pointer return data (0x700h) is subtracted and the value is used. Thus, the audio decoded data read pointer circulates in the storage area of the external storage device 100 to which the audio decoded data read pointer is assigned.

The update method of the audio decoded data write pointer is the same as that of the audio decoded data read pointer, and the write data amount is added to the audio decoded data write pointer of the corresponding channel, and the audio decoded data storage area of the corresponding channel is updated. The address data is compared, and only when the added audio decoded data write pointer is larger or equal, the audio decoded data pointer return data of the corresponding channel is subtracted, whereby the audio decoded data write pointer is assigned. External storage device 10
It goes around the storage area of 0.

The initial values of the audio decoded data read pointer and the audio decoded data write pointer are as follows:
Since it can be set arbitrarily, it is possible to make the area where the audio decoded data created by the IMDCT is written and the area where the audio decoded data necessary for the downmix operation processing is read different.

For each of the other channels 1 to 5, similarly to channel 0, an audio decoded data read pointer, audio decoded data write pointer, audio decoded data storage area final address data, and audio decoded data pointer return data are defined. As a result, the audio decoded data of each of the channels 1 to 5 is written or read.

FIG. 5 is a flowchart showing a control method for decoding the audio data of each block according to the present embodiment. In the middle of the decoding processing of the second block, the audio decoded data of each channel of the first block is transmitted a plurality of times. And mix down.

First, the register, the internal storage device 107,
Initialize the external storage device 100 (step S1
1). Then, a code data receiving process for inputting a bit stream already stored in the external storage device 100 is performed (step S12).

The bit stream is parsed to extract each exponent part Exp of the predetermined channel of the second block (step S13), and exponent part decoding processing for generating exponent data in the frequency domain using the extracted data. Is performed (step S14). Based on the generated exponent data in the frequency domain and each mantissa part Mant of the predetermined channel read from the bit stream, a mantissa bit allocation amount is calculated (step S15). Using this mantissa bit allocation, mantissa data in the frequency domain is generated (step S16).

Next, for audio decoded data of each channel of the first block already converted to the time domain and stored in the external storage device 100, the N times of downmix calculation processing performed by dividing into N is performed. It is determined whether or not the downmix operation has been completed (step S1). If not completed N times, the audio decoding data of each channel of the 1 / N block is read from the external storage device 100. At this time, by referring to the audio decoded data read pointer for each channel, while performing the read process from the external storage device 100, the read data amount is added to the audio decoded data read pointer, and the audio decoded data read pointer is added. Is compared with the audio decoded data storage area final address data. When the audio decoded data read pointer is greater than or equal to the audio decoded data storage area final address data, an update process of subtracting the audio decoded data pointer return data from the audio decoded data read pointer is performed (step S2).

Next, a downmix operation is performed to calculate 1 / N block of PCM data from audio decoded data of each channel of 1 / N block (step S3). The 1 / N block of PCM data is stored in the 1 / N block of the PCM data storage area 200 of the external storage device 100. At this time, the PCM data storage area 20
The write process is executed with reference to the PCM data write pointer indicating the head address of 0, and the write data amount is added to the PCM data write pointer (step S).
4).

If it is determined in step S1 that the processing has been completed N times, steps S2, S3, and S4 are not executed.

Next, decoded data in the frequency domain is formed from the mantissa data and the exponent data in the frequency domain of the second block, and the frequency domain is converted from the time domain to the time domain to generate audio decoded data in the time domain. Data synthesis and IMDCT are performed (step S17). And
The generated audio decoded data is stored in the external storage device 100.
At this time, while performing a write process on the audio decoded data write pointer indicating the head address of the storage area, the write data amount is added to the audio decoded data write pointer,
The audio decoded data write pointer is compared with the audio decoded data storage area final address data. If the audio decoded data write pointer is equal to or larger than the audio decoded data storage area final address data, an update process of subtracting the audio decoded data pointer return data from the audio decoded data write pointer is performed (step S18).

The processing of S12 to S18 and S2 to S4 is as follows.
It is repeated for each channel of the second block, each time the audio code data of each channel of the second block is converted to audio decoded data, and the audio decoded data of each channel of the first block is reduced by 1 / N. The data is mixed and converted into PCM data (step S19).

Next, it is determined whether or not the downmix operation processing of each channel of the first block has been performed N times (step S5). This is because the number of channels in the second block is compared with N, and if the number of channels in the second block is equal to or larger than N, the downmix calculation processing has already been completed. If the number of each channel is smaller than N, the downmix calculation process is not completed, and an unprocessed portion remains in each channel of the first block.

The unprocessed portion of each channel of the first block is processed in steps S6 to S8. That is, 1 /
The audio decoded data of each channel of N blocks is read from the external storage device 100 (step S6), and 1 /
A downmix operation is performed to calculate 1 / N blocks of PCM data from the audio decoded data of each channel of N blocks (step S3), and the 1 / N blocks of PCM data are stored in the PCM data storage area 200 of the external storage device 100. (Step S8).

As described above, according to the present embodiment, the audio decode data write pointer corresponding to each channel, the audio decode data read pointer corresponding to each channel, the PCM write pointer, and the audio decode data corresponding to each channel A data storage area final address data, audio decoded data pointer return data, one block of PCM data storage area 200, and one or more blocks of audio decoded data storage area corresponding to each channel are provided. Based on such a configuration,
The audio decoded data of each channel of the first block can be divided into N times and downmixed, and the encoded audio data of each channel of the second block can be decoded.

Therefore, the amount of data transferred at one time between the integrated semiconductor device 108 and the external storage device 100 can be reduced, and the memory bus can be used efficiently.

The present invention is not limited to the above embodiment, but can be variously modified. For example, it is possible to arbitrarily set the format of the bit stream, the encoded data, the decoded data, and the number of channels.

For example, in the above embodiment, one block is formed from a bit stream conforming to the AC-3 standard, that is, 6 channels (maximum) × 256 data (maximum).
Although a bit stream composed of a plurality of blocks is illustrated, the present invention is also applied to a bit stream conforming to another standard, for example, a bit stream including a frame composed of 8 channels (maximum) × 1024 data (maximum). be able to. In this case, the processing is performed by replacing the blocks of the above embodiment with frames. Further, the present invention can be applied even when the number of channels, the number of data, the number of blocks, and the like change dynamically. Or,
The present invention is not limited to a maximum of 50 subbands, and can be applied to any number of subbands.

(Embodiment 2) A signal processing apparatus according to Embodiment 2 of the present invention will be described with reference to FIGS. 9 to 11 and FIG.

First, the flow of the signal processing device of the present embodiment will be described with reference to FIG.

The bit stream input to the signal processing device 709 is a bit stream formed by encoding each of multi-channel audio signals into a frequency domain, and then expressing them by a mantissa part and an exponent part. Contains audio data for the channel.

Among the channels, there is a channel called a normal channel, which includes a high frequency band and a low frequency band. In addition, among the channels, there is a channel called a basic channel, which includes a high frequency band and a low frequency band shared by some other channels (channel 1400 shown in FIG. 21). In addition, among the channels, there is a channel called a coupling channel, which is configured by separating a high frequency band in advance and including only a low frequency band.

The bit stream is transmitted to the signal processing device 7
09 is stored in the storage device 700 outside the storage device 09. The bit stream parser 701 analyzes the syntax of the bit stream in the external storage device 700, extracts data such as the exponent part of each channel necessary for the decoding process, and extracts the extracted data in the signal processing device 709. Internal storage device 702
To be stored. The analyzed bit stream may be discarded from the external storage device 700.

Next, a decoding process for forming audio decoded data is performed for each channel included in the bit stream in the following procedure.

The exponent part decoder 703 in the frequency domain extracts the exponent part of each channel included in the bit stream from the internal storage device 702, decodes these exponent parts and generates each exponent data. Is stored in the work area in the internal storage device 702.

The mantissa data bit allocator 704 outputs the exponent data of the processing target channel generated by the frequency domain exponent part decoder 703 and the processing target channel included in the bit stream in the external storage device 700. Based on a mantissa indicating bit allocation, a bit allocation amount based on auditory characteristics is generated. Based on the bit allocation amount, mantissa data is generated by the mantissa decoder 705 in the frequency domain.

The generation of the mantissa data is performed according to the flowchart of FIG. First, the mantissa decoder 705
Determines whether the channel to be processed is a combined channel in which the high frequency band is separated in advance (step 20), and if the channel is the corresponding channel (step 20, YES), the low frequency band code of the channel is determined. The mantissa part of the data is decoded to form mantissa data, and this mantissa data is stored in a work area in the internal storage device 702,
The process proceeds to the frequency domain data generator 706 (step 24). The mantissa data of the low frequency band of the channel written in step 24 is stored in step 31 described later.
May be deleted as soon as the combining of the mantissa data and the exponent data of the channel is completed.

If it is not the corresponding channel (step 20, NO), the process proceeds to step 21 (step 2).
0). In this case, the channel to be processed is a normal channel composed of a high frequency band and a low frequency band, or a basic channel composed of a high frequency band and a low frequency band shared by some other channels. is there. For such a channel, the mantissa of the low frequency band and the mantissa of the high frequency band are decoded to obtain mantissa data of the low frequency band and mantissa data of the high frequency band, and these mantissa data are stored in the internal storage device 702. It is stored in the work area (step 21).

Next, it is determined whether or not the channel is a basic channel composed of a high frequency band and a low frequency band shared by several other channels (step 2).
2) If it is not the corresponding channel (step 22,
NO), proceed to the processing of frequency domain data generator 706; If the channel is applicable (step 22, YE
S), proceed to step 23; In step 23,
The mantissa data of the high frequency band of the basic channel generated in step 21 is written again to the area in the internal storage device 702.

In steps 21 and 23, the mantissa data of the high frequency band of the basic channel is stored in the internal storage device 70.
2 are written in two different areas. Since these areas are different from each other and are distinguished, two identical mantissa data are held in the internal storage device 702.

The mantissa data of the high frequency band of the basic channel written in step 21 may be deleted immediately after the combining of the mantissa data and exponent data of the basic channel is completed in step 31 described later.
The mantissa data of the high frequency band of the basic channel written in step 23 is stored in the internal storage device 70 until all the synthesis of the mantissa data and exponent data of each channel sharing the high frequency band code data of the basic channel is completed.
2.

The frequency domain data generator 706 combines the exponent data and the mantissa data generated by the exponent part decoder 703 and the mantissa part decoder 705 to generate decoded data in the frequency domain.

The generation of the decoded data in the frequency domain is performed according to the flowchart of FIG. First, it is determined whether or not the channel is a coupled channel in which a high-frequency band is separated in advance (step 30). If the channel is not a corresponding channel (step 30, NO), that is, if it is a normal channel or a basic channel, the internal storage device 702 It combines the exponent data and the mantissa data of the low frequency band and the exponent data and the mantissa data of the high frequency band to generate decoded data combining the low frequency band and the high frequency band (step 31).

If the channel is applicable (step 30, YES), the exponent data of the low frequency band and the mantissa data of the combined channel are combined, and the exponent data of the high frequency band of the basic channel and the basic channel obtained in step 21 are obtained. High-frequency mantissa data in the internal storage device 70
2 (step 32), the exponent data and the mantissa data of the high frequency band of the basic channel are combined (step 33), and decoded data combining the low frequency band and the high frequency band is generated (step 31).

The time domain transformer 707 transforms the frequency domain decoded data generated by the frequency domain data generator 706 into a time domain to generate a PCM signal.

In this embodiment, the decoding process of the mantissa part of the high frequency band common to a plurality of channels is performed in step 21 in FIG. 10, and the mantissa data obtained by this is stored in the internal storage device 702 in step 23. When storing and performing decoding processing for each channel, the mantissa data is repeatedly read from the internal storage device 702 and used.
For this reason, in order to obtain high-frequency band code data, access to the bit stream in the external storage device 700 is performed only once. No need to repeat it. Therefore, it is not necessary to hold the bit stream in the storage device until the decoding processing of the data in the high frequency band is completed, and a large storage capacity is not required. Further, once the decoding process of the data in the high frequency band is performed, it is not necessary to repeat this process again, so that the total amount of calculation can be reduced.

Further, since access to the bit stream in the external storage device 700 is performed only once in order to obtain high-frequency band code data, the bit stream 1100 in the external storage device 700 is accessed as shown in FIG.
The operation of the pointer 1103 for reading out the encoded exponent data and the encoded mantissa data from is simple as indicated by an arrow 1104. That is, since it is not necessary to repeatedly read the code data of the same high-frequency band included in the bit stream and only read the code data of each channel included in the bit stream in the order of the respective channels, the operation of the pointer 1103 is It is simple and the control of the pointer 1103 is simple.

(Embodiment 3) A signal processing apparatus according to Embodiment 3 of the present invention will be described with reference to FIGS.

The signal processing apparatus according to the third embodiment has the same configuration as the apparatus according to the second embodiment shown in FIG. 9, and performs the processing shown in FIG. 12 instead of the processing shown in FIG. The processing of FIG. 11 is performed in the same manner as the apparatus of the first embodiment. Therefore,
The signal processing device according to the third embodiment differs from the device according to the first embodiment only in performing the processing in FIG. 12 instead of the processing in FIG.

In the flowchart of FIG. 12, FIG.
Step 41 is inserted between steps 22 and 23 in the flowchart of FIG. That is, the mantissa decoder 705 determines whether or not the channel to be processed is a combined channel in which the high-frequency band has been separated in advance (step 20). If the channel is the corresponding channel (step 20, YES), The mantissa part of the low-frequency band code data of the channel is decoded to form mantissa data, the mantissa data is stored in a work area in the internal storage device 702, and the process proceeds to the frequency domain data generator 706 (step). 24).

If it is not the corresponding channel (step 20, NO), the process proceeds to step 21 (step 2).
0). In this case, the channel to be processed is a normal channel composed of a high frequency band and a low frequency band, or a basic channel composed of a high frequency band and a low frequency band shared by some other channels. is there. For such a channel, the mantissa of the low frequency band and the mantissa of the high frequency band are decoded to obtain mantissa data of the low frequency band and mantissa data of the high frequency band, and these mantissa data are stored in the internal storage device 702. It is stored in the work area (step 21).

Next, it is determined whether or not the channel is a basic channel composed of a high frequency band and a low frequency band shared by several other channels (step 2).
2) If it is not the corresponding channel (step 22,
NO), proceed to the processing of frequency domain data generator 706; If the channel is applicable (step 22, YE
S), the mantissa data of the high frequency band of the basic channel generated in step 21 is data-compressed (step 41), and the compressed mantissa data of the high frequency band is stored in the internal storage device 7;
02 is written again to the area within the area 02 (step 23).

The area in the internal storage device 702 where the mantissa data is written in step 21 and the area in the internal storage device 702 where the mantissa data after data compression is written in step 23 are different from each other and are distinguished. .

The mantissa data written in step 21 may be deleted immediately after the combining of the mantissa data and the exponent data of the basic channel is completed in step 31 which will be described later. The mantissa data of the band is held in the internal storage device 702 until the synthesis of the mantissa data and the exponent data of each channel sharing the high-frequency band code data of the basic channel is completed.

The frequency domain data generator 706 operates as shown in FIG.
The exponent data and the mantissa data generated by the exponent part decoder 703 and the mantissa part decoder 705 are combined in accordance with the flowchart of (1) to generate decoded data in the frequency domain. On this occasion,
After data compression in step 41, step 2
The mantissa data of the high frequency band of the basic channel written in 3 is stored in the internal storage device 7 every time the mantissa data and exponent data of each channel sharing the high frequency band are combined.
02 and expanded to the original mantissa data, and the original mantissa data is used.

In the present embodiment, since the mantissa data of the high frequency band is stored in the internal storage device 702 after the data is compressed in step 41, the capacity of the internal storage device 702 can be reduced.

(Embodiment 4) A signal processing apparatus according to Embodiment 4 of the present invention will be described with reference to FIGS.

The signal processing device of the fourth embodiment has the same configuration as the device of the second embodiment shown in FIG.
3 and the processing of FIG.

In the signal processing device of the fourth embodiment,
As in the first embodiment, the decoding process of the mantissa part of each channel is performed according to the flowchart in FIG. 10, and thus the description of the decoding process is omitted.

In the apparatus according to the second embodiment shown in FIG. 9, before performing the processing shown in FIGS. 10 and 11, the exponent part of each channel to be processed is extracted and decoded, and each exponent data is generated. Although stored in the work area in the internal storage device 702, in the signal processing device of the fourth embodiment,
Before performing the processing of FIGS. 13 and 14, the exponent part of each channel is not decoded, and the exponent part is decoded by the processing of FIG.

The generation of exponent data by decoding the exponent will be described with reference to the flowchart of FIG.

First, the exponent part decoder 703 determines whether or not the processing target channel is a combined channel in which a high frequency band is separated in advance (step 60). YE
S), the exponent part of the low-frequency band code data of the channel is decoded to form exponent data, and this exponent data is stored in a work area in the internal storage device 702, and is processed by the frequency domain data generator 706. Move on (step 64).

If it is not the corresponding channel (step 60, NO), the process proceeds to step 61 (step 6).
0). In this case, it is a normal channel composed of a high frequency band and a low frequency band, or a basic channel composed of a high frequency band and a low frequency band shared by some other channels. Also for such a channel, the exponent part of the low frequency band and the exponent part of the high frequency band are decoded to obtain the exponent data of the low frequency band and the exponent data of the high frequency band, and these exponent data are stored in the internal storage device 702. It is stored in the work area (step 61).

Next, it is determined whether or not the channel is the basic channel (step 62). If the channel is not the corresponding channel (step 62, NO), the frequency domain data generator 706 is determined.
Move on to processing. If the channel is applicable (step 62, YES), the process proceeds to step 63. Step 6
In step 3, the index data of the high frequency band of the basic channel generated in step 61 is written again to the area in the internal storage device 702.

In each of the steps 61 and 63, the index data of the high frequency band of the basic channel is stored in the internal storage device 70.
2 are written in two different areas. Since these areas are different from each other and are distinguished, two identical index data are held in the internal storage device 702.

The index data of the high frequency band of the basic channel written in step 61 may be deleted immediately after the combining of the mantissa data and the exponent data of the basic channel is completed in step 71 described later.
The exponent data of the high frequency band of the basic channel written in step 63 is stored in the internal storage device 70 until all the synthesis of the mantissa data and the exponent data of each channel sharing the high frequency band code data of the basic channel is completed.
2.

Next, the generation of decoded data in the frequency domain by the frequency domain data generator 706 will be described with reference to the flowchart of FIG.

First, the frequency domain data generator 706
It is determined whether or not the processing target channel is a combined channel in which a high-frequency band is separated in advance (step 70).
If it is not the corresponding channel (step 70, N
O), that is, if the processing target channel is a normal channel or a basic channel, the internal storage device 7
The exponent data in 02 and the mantissa data in step 21 are combined to generate frequency-domain decoded data (step 71). That is, the exponent data and the mantissa data of the low frequency band are combined, and the exponent data and the mantissa data of the high frequency band are combined to generate decoded data in which the low frequency band and the high frequency band are combined.

If the channel is applicable (step 70, YES), the exponent data and the mantissa data of the low frequency band of the combined channel are synthesized, and the mantissa data of the high frequency band of the basic channel in step 21 and the data in step 61. The obtained index data of the high frequency band of the basic channel is retrieved from the internal storage device 702 (step 7).
2) Combine exponent data and mantissa data of the high frequency band of the basic channel (step 73), and generate decoded data combining the low frequency band and the high frequency band (step 7).
1).

The time domain converter 707 converts the frequency domain decoded data generated by the frequency domain data converter 706 into the time domain to generate a PCM signal.

In this embodiment, the decoding process of the mantissa of the high frequency band common to a plurality of channels is performed in step 21.
After performing in step 61, not only the mantissa data obtained in this way is stored in the internal storage device 702 in step 23, but also the exponent part of the high frequency band common to the respective channels is decoded in step 61. The exponent data obtained in step 63 is stored in the internal storage device 702 in step 63, and the mantissa data and the exponent data are repeatedly read from the internal storage device 702 and used when decoding the channels. As a result, the processing amount can be reduced.

Embodiment 5 A signal processing apparatus according to Embodiment 5 of the present invention will be described with reference to FIGS.

The signal processing device of the fifth embodiment has the same configuration as the device of the first embodiment shown in FIG.
7 is performed.

In the signal processing device of the fifth embodiment,
As in the fourth embodiment, the decoding process of the exponent part of each channel is performed according to the flowchart of FIG.

In the signal processing device of the fifth embodiment, the decoding process of the mantissa part of each channel is performed according to the flowcharts of FIGS. 15 and 16, and the combining process of the mantissa data and the exponent data is performed as shown in FIG. This is performed according to the flowchart.

First, the mantissa data bit allocator 704
Will be described with reference to the flowchart of FIG.

The mantissa data bit allocator 704 determines whether the channel to be processed is a combined channel in which a high frequency band has been separated in advance (step 8).
0), if it is the corresponding channel (step 80, YES), the power spectrum density is obtained from the index data of the low frequency band of the channel, and the bit allocation amount based on the auditory characteristics is calculated (step 84). The process proceeds to the processing of the partial decoder 705.

If it is not the corresponding channel (step 80, NO), the process proceeds to step 81 (step 80). In this case, the channel to be processed is a normal channel composed of a high frequency band and a low frequency band, or a basic channel composed of a high frequency band and a low frequency band shared by some other channels. is there. For such a channel, the power spectrum density is obtained from the decoded low frequency band and high frequency band index data of the channel, and the bit allocation amount of the low frequency band and the high frequency band based on the auditory characteristics is calculated (step 81). .

Next, it is determined whether or not the channel to be processed is the basic channel (step 82). If the channel is not the corresponding channel (step 82, NO), the process proceeds to the mantissa decoder 705. If the channel is applicable (step 82, YES), the process proceeds to step 83. In step 83, the bit allocation amount of the high frequency band of the basic channel generated in step 81 is written in the internal storage device 702 (step 83), and the process proceeds to the mantissa decoder 705.

The bit allocation amount of the high frequency band of the basic channel written in step 63 is stored in the internal storage device 702 until all the mantissa data and exponent data of each channel sharing the high frequency band of the basic channel are completed. Will be retained.

Next, the processing by the mantissa decoder 705 will be described with reference to the flowchart of FIG.

The mantissa decoder 705 determines whether the channel to be processed is a combined channel in which the high-frequency band has been separated in advance (step 90), and if the channel is the corresponding channel (step 90, YES). ,
The bit allocation amount of the high frequency band of the basic channel stored in step 83 is read from the internal storage device 702 (step 92), and based on the bit allocation amount of the high frequency band, mantissa data of the high frequency band of the basic channel is generated and The mantissa data of the low frequency band of the combined channel is generated based on the bit allocation amount of the low frequency band of the combined channel generated in step 84 (step 91). After that, the processing moves to the frequency domain data generator 706.

If the channel is not the corresponding channel (step 90, NO), the channel to be processed is a normal channel composed of a high frequency band and a low frequency band, or is shared by some other channels. This is a basic channel composed of a high frequency band and a low frequency band. For such a channel, mantissa data of the channel to be processed is generated based on the bit allocation amount generated in step 81, and the process proceeds to the frequency domain data generator 706 (step 9).
1).

Next, the generation of decoded data in the frequency domain by the frequency domain data generator 706 will be described with reference to the flowchart of FIG.

First, the frequency domain data generator 706
It is determined whether or not the processing target channel is a combined channel in which a high frequency band is separated in advance (step 100).
0), if it is not the corresponding channel (step 1)
000, NO), that is, if the channel is a normal channel or a basic channel, the exponent data of the channel to be processed is read from the internal storage device 702, and the exponent data is combined with the mantissa data of the channel to be processed from the mantissa decoder 705. Then, decoded data in the frequency domain is generated (step 1001).

If the channel is applicable (step 1000, NO), the index data of the high frequency band of the basic channel is read from the internal storage device 702 (step 1).
002). Further, the exponent data of the low frequency band of the coupled channel and the mantissa data are synthesized, and the mantissa data of the high frequency band of the basic channel obtained in step 91 and the exponent data of the high frequency band of the basic channel from the internal storage device 702 are synthesized. (Step 1003) Generate decoded data combining the low frequency band and the high frequency band (Step 1)
001).

The time domain converter 707 converts the frequency domain decoded data generated by the frequency domain data generator 706 into a time domain to generate a PCM signal.

In the present embodiment, when performing decoding processing of a combined channel in which a high-frequency band is separated in advance,
Exponent part decoder 703 and mantissa data bit allocator 7
04 can be omitted. Further, the data size of the data indicating the bit allocation amount is about １ ／ as compared with the size of the mantissa data, so that the capacity of the internal storage device 702 can be reduced. As a result, when decoding processing of a combined channel in which a high-frequency band is separated in advance, processing can be speeded up.

Further, it is not necessary to hold the bit stream in the storage device until the decoding process of the code data in the high frequency band is completed, so that a large storage capacity is not required.

When decoding the combined channel in which the high-frequency band has been separated in advance from the combination of each of the second to fifth embodiments, only the processing of the exponent part decoder 703 in the frequency domain is omitted, or the mantissa data bit allocation is performed. Vessel 704
Or only the processing of (2) is omitted, and the mantissa decoder 7 in the frequency domain
It is possible to omit only the process 05 and thereby speed up the entire process.

[0153]

As described above, according to the audio decoder apparatus of the present invention, the audio decoded data write pointer corresponding to each channel, the audio decoded data read pointer corresponding to each channel, the PCM write pointer, By providing the final address data of the audio decoded data storage area corresponding to the channel, the return data of the audio decoded data pointer, one block of PCM data storage area, and one or more blocks of audio decoded data storage area corresponding to each channel, The control of dividing the block data into N times to execute the downmix operation processing and the control of executing the processing using the audio block data immediately before the audio block data currently being processed can be realized. Day with equipment Can be reduced transfer amount, it is possible to use the memory bus efficiency.

According to the signal processing apparatus of the present invention, a bit stream including code data of a plurality of channels is input, and common code data common to each channel included in at least one of the channels is decoded. To form common decoded data, and for each of the channels, decode channel code data unique to the channel to form channel decoded data, and combine the channel decoded data with the common decoded data, thereby In a signal processing device for forming each channel decoded data,
When performing decoding processing of each channel sharing the common code data, the processing speed can be increased, and the processing form does not need to hold a bit stream.
Decoding processing can be performed even in a signal processing device that cannot perform a bit stream holding operation.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating an audio decoder device according to a first embodiment of the present invention.

FIG. 2 is a diagram showing a configuration of a bit stream in the audio decoder device of FIG.

FIG. 3 is a diagram showing memory mapping of an external storage device in the audio decoder device of FIG. 1;

FIG. 4 is a diagram for explaining an external storage device access method in the audio decoder device of FIG. 1;

FIG. 5 is a flowchart illustrating an operation of the audio decoder device of FIG. 1;

FIG. 6 is a block diagram showing a conventional audio decoder device.

FIG. 7 is a diagram showing memory mapping of an external storage device in the conventional device of FIG. 6;

FIG. 8 is a flowchart illustrating the operation of the conventional device of FIG. 6;

FIG. 9 is a block diagram illustrating a signal processing device according to a second embodiment of the present invention.

FIG. 10 is a flowchart illustrating an operation of a mantissa decoder in the signal processing device according to the second embodiment.

FIG. 11 is a flowchart illustrating an operation of a frequency domain data generator in the signal processing device according to the second embodiment.

FIG. 12 is a flowchart for explaining another operation of the mantissa decoder in the signal processing device of the third embodiment.

FIG. 13 is a flowchart illustrating an operation of an exponent part decoder in the signal processing device according to the fourth embodiment.

FIG. 14 is a flowchart illustrating an operation of a frequency domain data generator in the signal processing device according to the fourth embodiment.

FIG. 15 is a flowchart illustrating an operation of a mantissa data bit allocator in a signal processing device according to a fifth embodiment.

FIG. 16 is a flowchart illustrating the operation of a mantissa decoder in the signal processing device according to the fifth embodiment.

FIG. 17 is a flowchart illustrating an operation of a frequency domain data generator in the signal processing device according to the fifth embodiment.

FIG. 18 is a diagram illustrating movement of a pointer in the signal processing device of the present invention for accessing a bit stream.

FIG. 19 is a diagram showing the movement of a pointer of a conventional device for accessing a bit stream.

FIG. 20 is a block diagram showing a conventional signal processing device.

FIG. 21 is a diagram for describing channels that combine high-frequency bands in the signal processing device.

[Explanation of symbols]

100, 500 External storage device 101, 501 Input bit stream parser 102, 502 Exponent decoder 103, 503 Mantissa data bit allocator 104, 504 Mantissa decoder 105, 505 IMDCT unit 106, 506 Downmix calculator 107 , 507 Internal storage device 108, 508 Integrated semiconductor device 200, 600 PCM data storage area (1 block) 201, 301 Channel 0 audio decoded data (1.75 blocks) 202, 302 Channel 1 audio decoded data (1 .75 blocks) 203 Audio decoded data of channel 2 (2.7
(5 blocks) 204 Channel 3 audio decoded data (4.7
205 audio decoded data of channel 4 (4 blocks) 206 audio decoded data of channel 5 (1.5 blocks)
Block) 601 Channel 0 audio decoded data (1 block) 602 Channel 1 audio decoded data (1 block) 603 Channel 2 audio decoded data (1 block) 604 Channel 3 audio decoded data (1 block) 605 Channel 4 audio decoded data Audio decoded data (1 block) 606 Audio decoded data of channel 5 (1 block) 700 External storage device of signal processing device 701, 1300 Input stream parser 702, 1301 Internal storage device of signal processing device 703, 1302 Frequency domain Exponent decoder 704, 1303 Mantissa data bit allocator 705, 1304 Frequency domain mantissa decoder 706, 1305 Frequency domain data generator 707, 1306 Time domain converter 709 Signal Processing equipment

Continued on the front page (72) Inventor Masahiro Sueyoshi 1006 Kazuma Kadoma, Osaka Pref. Matsushita Electric Industrial Co., Ltd. (72) Inventor Takeshi Fujita 1006 Odaka Kazuma Kadoma, Osaka Pref. Takashi Katayama 1006 Kadoma, Kazuma, Kadoma, Osaka Prefecture, Japan Matsushita Electric Industrial Co., Ltd. (72) Inventor Kazuma Abe 1006, Kadoma, Kadoma, Kadoma, Osaka Prefecture Inside Matsushita Electric Industrial Co., Ltd. 1006 Kadoma Matsushita Electric Industrial Co., Ltd. (72) Inventor Eiji Otomura 1006 Kadoma, Kadoma-shi, Osaka Prefecture Matsushita Electric Industrial Co., Ltd. (72) Inventor Akihisa Kawamura 1006 Kadoma, Kadoma, Kadoma-shi, Osaka Matsushita Electric Industrial Co., Ltd. In company

Claims (17)

[Claims] 1. A bit stream is input in units of blocks, a bit stream of one block is decoded to form audio decoded data of a plurality of channels, and the audio decoded data of each channel is stored in storage means. In the audio decoder device for downmixing the audio decoded data of each channel in the storage means, when decoding the bit stream of the second block, the decoding of the channel corresponding to the bit stream of the first block in the storage means is performed. An audio decoder device comprising an arithmetic unit for downmixing audio decoded data. 2. The bit stream of the second block is converted into audio decoded data of each channel by a plurality of decoding processes, and the arithmetic unit executes the decoding process in the storage unit each time the decoding process is performed. 2. The audio decoder device according to claim 1, wherein the audio decoded data of each channel corresponding to one block of the bit stream is divided and sequentially downmixed. 3. The bit stream of the second block is converted into audio decoded data for each of the channels by repeating a decoding process by the number of each of the channels. At each processing,
2. The audio decoder device according to claim 1, wherein the audio decoding data of each channel corresponding to the bit stream of the first block in the storage unit is divided and sequentially downmixed. 4. The audio decoder device according to claim 1, wherein the audio decoding data formed by the downmixing is temporarily stored in the storage unit and then output. 5. An audio decoder device which converts a multi-channel audio signal into a frequency domain and then inputs and decodes a bit stream formed by encoding to be represented by a mantissa part and an exponent part. A bitstream parser for parsing the stream and extracting data necessary for decoding processing; an internal storage device for storing data necessary for the decoding process; and the audio based on data in the internal storage device. An exponent part decoder that generates exponent data in the frequency domain of the signal; a mantissa data bit allocator that calculates a mantissa bit allocation amount from the exponent data output from the exponent part decoder; and an output from the mantissa data bit allocator. Generating mantissa data in the frequency domain of the audio signal based on the assigned amount Mantissa decoder, and performing exponential data and mantissa data generated by the exponent decoder and the mantissa decoder from the frequency domain to the time domain, thereby performing audio decoding for each channel. An IMDCT unit for forming data; and a downmix calculator for generating and interleaving PCM data from the audio decoded data of each channel, and storing the bit stream, the audio decoded data and the PCM data in an external storage device. When the bit stream is input block by block and the bit stream of the second block is decoded, the PCM is decoded from the audio decoded data of each channel corresponding to the bit stream of the first block already stored in the external storage device. Audio decoder that generates data Apparatus. 6. The external storage device includes a PCM data storage area and an audio decoding data storage area corresponding to each channel, wherein the PCM data storage area includes a data amount of a plurality of channels × a plurality of data. A capacity capable of storing PCM data corresponding to a block, wherein the audio decoded data storage area is divided into respective areas corresponding to respective channels, and any area exceeds the one block of the corresponding channel. 6. A capacity capable of storing decoded data.
An audio decoder device according to claim 1. 7. An audio decoding data write pointer corresponding to each channel for writing the audio decoding data to the external storage device, and an audio decoding corresponding to each channel for reading the audio decoding data from the external storage device. A data read pointer, a PCM write pointer for writing the PCM data to the external storage device, the audio decode write pointer, an audio decode data storage area address data corresponding to each channel for updating the audio read pointer, and Audio decoded data pointer return data, wherein the audio decoded data write pointer and the audio decoded data read pointer are updated independently, and request to circulate in an area allocated to each channel. Section 5
An audio decoder device according to claim 1. 8. The audio decoder device according to claim 5, wherein the downmix calculator executes processing by dividing the audio decoded data of each channel into N times. 9. A bit stream including code data of a plurality of channels is input, and common code data common to each channel included in at least one of the channels is decoded to form common decoded data. For each channel, signal processing for decoding channel code data unique to the channel to form channel decoded data, and combining the channel decoded data with the common decoded data, thereby forming decoded data for each channel. In the device, storage means for storing common decoded data formed by decoding the common code data, and each time the channel code data unique to the channel is decoded to form channel decoded data, the storage means The common decrypted data is read from the Signal processing device and a control means for causing the binding of the decoded data. 10. A bit stream including code data of a plurality of channels is input, and common code data common to each channel included in at least one of the channels is decoded to form common decoded data. For each channel, signal processing for decoding channel code data unique to the channel to form channel decoded data, and combining the channel decoded data with the common decoded data, thereby forming decoded data for each channel. In the apparatus, storage means for storing intermediate data in the decoding processing of the common code data, and each time decoding processing of channel code data unique to the channel to form channel decoded data, the storage means stores the intermediate data from the storage means. Read and form the common decryption data from this intermediate data , The signal processing device and a control means for causing the coupling of the common decoded data and the channel decoded data. 11. A multi-channel audio signal is converted into a frequency domain, and then a bit stream formed by encoding to be represented by a mantissa and an exponent is input, and the bit stream included in at least one of a plurality of channels is input. The high frequency band code data common to each channel is decoded to form high frequency band decoded data, and the low frequency band code data is decoded for each channel to form low frequency band decoded data. A signal processing device for combining band-decoded data with the high-frequency band-decoded data, thereby forming decoded data for each of the channels; and an input stream parsing for parsing the bit stream and extracting data necessary for decoding processing. An internal storage device for storing data necessary for decoding processing; and a data storage device in the internal storage device. An exponent part decoder that generates exponent data in the frequency domain of the audio signal based on the data, a mantissa data bit allocator that calculates a mantissa bit allocation amount from the exponent data output from the exponent part decoder, A mantissa decoder that generates mantissa data in the frequency domain of the audio signal based on the allocation amount output from the mantissa data bit allocator; and an exponent generated by the exponent decoder and the mantissa decoder. From the data and the mantissa data, synthesize the high-frequency band decoded data and the low-frequency band decoded data of each channel, combine the low-frequency band decoded data of each channel with the high-frequency band decoded data, and shift from the frequency domain to the time domain. And a data generator for forming decoded data of each channel by performing the conversion of The waveband decoded data is stored in the internal storage device, and when forming the low frequency band code data of the channel, the high frequency band decoded data is read from the internal storage device, and the low frequency band code data is stored in the internal storage device. A signal processing device for coupling to high frequency band decoded data. 12. The signal processing device according to claim 11, wherein the high-frequency band decoded data is compressed and stored in the internal storage device. 13. A multi-channel audio signal is converted into a frequency domain, and then a bit stream formed by encoding represented by a mantissa part and an exponent part is input, and the bit stream included in at least one of a plurality of channels is input. The high frequency band code data common to each channel is decoded to form high frequency band decoded data, and the low frequency band code data is decoded for each channel to form low frequency band decoded data. A signal processing device for combining band-decoded data with the high-frequency band-decoded data, thereby forming decoded data for each of the channels; and an input stream parsing for parsing the bit stream and extracting data necessary for decoding processing. An internal storage device for storing data necessary for decoding processing; and a data storage device in the internal storage device. An exponent part decoder that generates exponent data in the frequency domain of the audio signal based on the data, a mantissa data bit allocator that calculates a mantissa bit allocation amount from the exponent data output from the exponent part decoder, Based on the mantissa bit allocation amount output from the mantissa data bit allocator, a mantissa decoder that generates mantissa data in the frequency domain of the audio signal, an exponent decoder and a mantissa decoder are generated by the mantissa decoder. The high frequency band decoded data and the low frequency band decoded data of each channel are combined from the exponent data and the mantissa data, and the low frequency band decoded data of each channel is combined with the high frequency band decoded data. A data generator that forms decoded data of each of the channels by performing conversion to a region. The intermediate data in the decoding process of the high-frequency band code data is stored in the internal storage device, and when forming the low-frequency band code data of the channel, the intermediate data is read from the internal storage device, and the intermediate data is read. A signal processing device that forms the high-frequency band decoded data from the data and combines the low-frequency band code data with the high-frequency band decoded data. 14. The signal processing device according to claim 13, wherein the intermediate data is compressed and stored in the internal storage device. 15. The signal processing apparatus according to claim 13, wherein the intermediate data is exponential data output from the exponent decoder. 16. The signal processing device according to claim 13, wherein the intermediate data is a mantissa bit allocation amount output from the mantissa data bit allocator. 17. The signal processing device according to claim 13, wherein the intermediate data is mantissa data in the frequency domain output from the mantissa decoder.