KR20060105931A - Decoding control method for audio frame - Google Patents

Abstract

The present invention provides a method for controlling decoding of an audio frame in which an audio frame, in particular, recognizes a low stream format and can decode the recognized low stream format using a sampling frequency.

The present invention provides an AAC-based audio data decoding method comprising: searching an audio stream; Checking the presence or absence of the stream header; Performing decoding in a corresponding format when the stream header is found; During the decoding, if the decoding is not normally performed, repeating the decoding process at a start position of the next frame a predetermined number of times, and recognizing it as a stream if the decoding is not normally performed; If it is recognized as the raw stream, characterized in that it comprises the step of performing decoding using the sampling frequency information.

AAC, ADIF, ADTS, Decryption, Row Stream

Description

Decoding control method for audio frame

1 illustrates an ADIF header structure.

2 illustrates an ADTS header structure.

3 is a flowchart illustrating a decoding control method of an audio frame according to an exemplary embodiment of the present invention.

4 illustrates a bit stream structure with an ADIF header.

5 illustrates a bit stream structure with an ADTS header.

6 is a diagram illustrating a one-channel AAC bit stream segment structure.

7 illustrates a two-channel AAC bit stream segment structure.

8 is a diagram showing a program example of a method for controlling decoding of an audio stream according to the present invention;

The present invention provides a method for controlling decoding of an audio frame in which an audio frame, in particular, recognizes a low stream format and can decode the recognized low stream format using a sampling frequency.

Digital audio, developed from the beginning of compact disks (CDs), has been developed to reduce storage capacity by compressing audio in order to efficiently transmit or store large amounts of data. Since 1992, the audio coding method has been internationally standardized by the Moving Picture Experts Group (MPEG). In addition, Dolby, Sony and AT & T have developed their own audio coding methods.

MP3, which is widely used at present, refers to MPEG-1, 2 audio layer 3. In general, MP3 has been developed to encode less data while maintaining sound quality similar to that of CD. These methods are preferentially adopted and used in a wireless environment that requires efficient use of the transmission band, and the range of their application is increasing.

A typical example is MPEG-2 / 4 AAC (Advanced Audio Coding), which is widely used in VOD (Video On Demand), AOD (Audio On Demand), etc., providing sound quality similar to MP3 128kbps with 96kbps of data. Recently, MP3 players have been introduced to support AAC as a multiple audio decoder.

The AAC file has the Audio Data Interchange Format (ADIF) and the Audio Data Transport Stream (ADTS) headers as interface formats. Depending on the encoder, the AAC file can also generate an AAC file in the form of a raw stream without these headers. . In general, since the table or frame boundary required for decoding can be set by using the header information at the beginning of the file, decoding is impossible if the information is not found. Especially, in the case of a low stream, there is no header information. As a result, playback in a general decoder is impossible.

AAC (Advanced Audio Coding), which is one of audio coding methods of MPEG, is an interface format, as shown in FIG. 1, and has an audio data interchange format (ADIF) structure, and FIG. 2 shows an audio data transport stream (ADTS) structure. It is shown.

In the case of the AAC audio content format, one audio content is composed of data sampled by a predetermined time unit and a header indicating audio information. This one block is called a frame, and a data block in each of the plurality of blocks is provided. Audio data exists in the raw_data_block, and includes all the information of the audio content in a header field corresponding to each data block raw_data_block.

As shown in FIG. 1, the ADIF header includes a header ID (adif_id), copyright-related information (copyright_id_present, copyright_id), information on whether or not the original is copied (original / copy), home-related information (home), and bit stream type information ( bit stream type, bit rate information, configuration information (num_program_config_element), buffer information (adif_buffer_fullness), program configuration information (program_config_element), and the like. Numerical values written in each field in the figure represent the number of bits allocated to the corresponding information.

On the other hand, as shown in FIG. 2, the ADTS header is divided into an invariant header (ADTS_fixed_header) whose value does not change for each frame and a variable header (ADTS_varialbe_header) that stores numerical values specific to each frame. The immutable header includes sync word information (sync word), ID information (ID), layer information (layer), protection information (protection_absent), profile information (profile), sampling frequency information (sampling_freq._index), and private information (private_bit). , Channel configuration information (channel_config.), Original / copy information (original / copy), home information (home) and the like. The variable header includes copyright related information (copyright_id_bit, copyright_id_start), frame length information (frame_length), buffer information (adts_buffer_fullness), information related to a raw data block in a frame (num._of_raw_data_blocks_in_frame), and the like. There is an error check field (ADTS_error_check) at the end of the ADTS header, and information for error check (crc_chek) is included.

Among the AAC interface formats, ADTS has a large overhead because each frame has a header.However, if an error occurs in the middle of the stream, the next frame can be found using the frame header. has the advantage of performing decoding. However, because ADIF has a header only at the head of the file, the overhead is small, but when an error occurs in the middle of the stream, it is difficult to find the start of the next frame.

Data delivered in streaming format, such as AOD (Audio On Demand) or broadcasting, has a great possibility of being lost and transmitted in the middle by various communication environments depending on the reception state of the radio wave or the state of the Internet network.

As a result, the audio frame data necessary for decoding may be partially damaged or lost by more than a few frames. Therefore, dealing with such loss of the audio data stream is a very important matter to consider and deal with.

As mentioned earlier, the ADTS format can find and decode the beginning of the next frame when an error occurs, but in the case of a raw stream that does not have the ADIF format or the ADTS and ADIF formats, the next frame can be used even if an error occurs. Since the start of is not found, this results in a state in which decoding can no longer be performed and the decoding process is terminated.

That is, in the conventional AAC-based audio decoding system, decoding is interrupted when the audio data stream is lost in the middle, which causes a significant drop in audio reproduction quality and a decrease in reliability of the decoding system.

According to the present invention, when the stream header does not come to the beginning of the file, the ACC player determines what format the stream has been received, and if the stream is received, what channel information and sampling frequency information has been received. The present invention provides a decoding control method of an audio frame to enable playback.

The present invention provides a method of controlling the decoding of an audio frame, by searching for an audio stream, determining whether the stream has an ADIF / ADTS header, and then decoding the stream without or without decoding the ADIF / ADTS header. In providing.

The present invention has a stream arrangement form identical to the ID of ADIF or the sync word of ADTS in a raw stream type AAC file. It is possible to find out, and to change the sampling frequency value when it turns out to be a low stream, to find and play back a frequency where decoding is normally performed.

Decoding control method of an audio frame of the present invention for achieving the above object,

In the AAC-based audio data decoding method,

Searching for the audio stream;

Checking the presence or absence of the stream header;

Performing decoding in a corresponding format when the stream header is found;

During the decoding, if the decoding is not normally performed, repeating the decoding process at a start position of the next frame a predetermined number of times, and recognizing it as a stream if the decoding is not normally performed;

If it is recognized as the raw stream, characterized in that it comprises the step of performing decoding using the sampling frequency information.

Preferably, the stream header is characterized in that the ADIF or ADTS header.

Preferably, if the stream header is not found, the stream is recognized as a low stream.

Specifically, the decoding of the raw stream may include finding a sampling frequency candidate group that is normally decoded with respect to a previous frame; And reproducing at a high sampling frequency among the candidate groups.

Specifically, when reproducing at the normal sampling rate, when reproducing at the normal speed, and if the reproducing is not performed at the normal speed, further comprising the step of finding and reproducing at a low sampling frequency from the candidate group until reproducing at the normal speed; Characterized in that.

Specifically, the method may further include determining whether to switch to a low sampling frequency after receiving confirmation from the user about whether to play at the normal speed.

According to another aspect of the present invention, there is provided a method for controlling decoding of an audio frame, the method comprising: analyzing a format of a file in which an ADIF or ADTS header does not appear at a file start position; As a result of the analysis, it is characterized in that it comprises the step of finding and reproducing the sampling frequency suitable for the reproduction speed while varying the sampling frequency in accordance with the sampling frequency group in the case of a low stream.

Specifically, decoding is performed while varying from the highest sampling frequency in the sampling frequency group to a lower sampling frequency in stages.

Hereinafter, exemplary embodiments of the present invention will be described with reference to the accompanying drawings.

In the present invention, the presence of an audio stream header, for example, an ADTS / ADIF header, is checked at the beginning of the file. In this case, when there is no ADTS / ADIF header, the AAC player checks which format the stream is received, and then decodes and reproduces the raw stream using channel information and sampling frequency.

In addition, the present invention is characterized in that it is possible to find and reproduce a frequency for which decoding is normally performed while changing the sampling frequency value when it is determined as a low stream.

Hereinafter, with reference to the accompanying drawings as follows.

3 is a flowchart illustrating an audio stream decoding method according to an exemplary embodiment of the present invention.

Referring to FIG. 3, the received AAC-based audio stream is searched (S101). At this time, it is checked whether the ADIF or ADTS header shown in FIGS. 1 and 2 is found at the start position of the received stream (S102). In other words, if the AAC file is input, the presence or absence of the ADIF / ADTS header is checked. The search is performed to determine whether there is a stream having the same array as the 32-bit ADIF ID or the 12-bit ADTS sync word. At this time, if there is a matching array, it is regarded as an ADIF or ADTS format, and the decryption is attempted by applying the syntax of the format.

As a result of checking in S102, if a header is found, decoding is regarded as an ADIF or ADTS format and attempted to decrypt (S103).

However, as a result of checking in S102, when the ADIF or ADTS header is not found at the stream start position or when normal decoding is not performed in S104, the first frame start position according to the boundary condition is found as a low stream.

4 shows a bit stream structure with an ADIF header, and FIG. 5 shows a bit stream structure with an ADTS header.

As shown in FIG. 4, the start of every frame in the ADIF or raw format starts with three bits of data called an element ID ele_id. This element ID (ele_id) represents IDs of elements that can be used in every frame. In a mobile AOD environment, ID_SCE [000], ID_CPE [001], ID_FIL [110], ID_END Four of [111] are mainly used.

Among them, ID_SCE [000] has data of one channel (mono), and ID_CPE [001] has data of two channels (stereo). The case of ID_FIL [110] includes data for adjusting bitrate or data for extended function such as AAC +. ID_END [111] is an element indicating the end of a frame.

The bit stream structure with this ADIF header has an ADIF header and consists of a block of data, element ID, within each frame.

Referring to the bit stream structure having the ADTS header shown in FIG. 5, each frame includes an ADTS synchronization, an ADTS header, an element ID, and a low data block.

6 and 7 illustrate various forms in which the elements constitute a frame. In general, the frame of a mono channel starts with ID_SCE [000] and ends with ID_FIL [110] as shown in FIG. 6, and the frame of a stereo channel starts with ID_CPE [001] and ID_FIL [110 as shown in FIG. 7. Ends with].

If the ADIF / ADTS header is not found in the confirmation result of S102 or if decoding is not normally performed when decoding of S104, the stream is regarded as a low stream and the first frame start position is found according to a boundary condition.

Here, the frame that is not normally decoded at the time of decoding in S104 is a low-stream AAC file, which has the same stream arrangement type as the ID of ADIF or the sync word of ADTS. In this case, even if it is found as an ADIF or ADTS header, it is wrongly determined and can be determined as a low data format.

Therefore, if an error occurs in the decoding process and the next frame start position is found, referring to the channel information of the previous normal decoded frame, that is, the element ID ele_id listed above, the start position of the next frame You can expect the element ID (ele_id) to come in. For example, if the channel information obtained from a normal decoded previous frame is ID_SCE [000], the next frame may be regarded as a monochannel frame, and if ID_CPE [001], the next frame may be regarded as a stereo channel frame. In the former case, the position where ID_SCE [000] appears is found. In the latter case, once the position where ID_CPE [001] appears, this position can be regarded as the start position of the next frame. However, because there may be similar bit arrays by chance, many different verification and verification processes are required for more reliable frame retrieval and recognition. Element instance tag (element_instance_tag) and ID_END [111] are used for such verification and confirmation.

Both ID_SCE [000] and ID_CPE [001] described above have 4-bit element instance tag (element_instance_tag) data. This is used to distinguish when several elements are used in the same frame at the same time. In general, an element at the beginning of the frame has an element instance tag (element_instance_tag) value of [0000].

The ID_END element [111] that ends each frame is followed by the beginning of the frame (ele_id indicating ID_SCE or ID_CPE) after byte alignment. Although it is not a clear distinction like the ADTS header, it is possible to find the frame boundary when an error occurs as described above by using the start and end data of this partial frame as an index.

Here, the start position of the next frame is based on the channel information obtained from a previously decoded frame, and the element ID (ele_id) value and the element instance tag (element_instance_tag) value to be expected at the start position of the next frame after the error is estimated If we find the bit stream of this emerging position, we can assume that the position is the start of the next frame.

As such, when the starting position of the first frame according to the boundary condition of the raw stream is found in S105, a candidate group of sampling frequencies normally decoded from the previous frame is found. In the candidate group, playback is performed while decoding at a high sampling frequency (S107).

At this time, after confirming that playback is performed at the normal speed (S108), if playback is not performed at the normal speed, the playback is performed while decoding at a low sampling frequency among the candidate groups (S109). In other words, while lowering the sampling frequency, decoding is performed until a normal reproduction speed is obtained. In this case, whether the normal playback speed is normal or not, the user checks whether the stream being played is normally output. If the playback speed is faster than the actual speed, the user inputs a pre-defined key input to switch to a lower sampling frequency value in the same sampling frequency group. Will be. As another example, when applying the sampling frequency may be applied while increasing the frequency in order from the lowest frequency, it may be applied by any order of application.

Specifically, the most important information in the ADIF / ADTS header is sampling frequency information, and the scale factor band table used in the AAC is changed according to this information. In this case, the number of scale factors and the offset value are also different, and as a result, the amount of data used in the decoding process is different. Therefore, when the wrong sampling frequency is set, decoding is not performed at all.

In contrast, the raw stream can be normally decoded if the sample can accurately grasp only the frequency information. For example, the channel information can be identified using an element eye value corresponding to the first 3 bits of the raw stream.

As the sampling frequency used in the AAC file, for example, 12 predefined frequencies in the range of 8 to 96 khz are used. The scale factor band tables share the same table among similar sampling frequencies. For example, in the case of a long window scale factor table, 96 / 88.1khz, 22.05 / 24khz, and 12 / 11.025 / 8khz use the same table. Therefore, in the case of the low stream AAC file, the same decoding result can be obtained in the sampling frequency group using the same table.

Using this feature, decoding is performed using the highest sampling frequency value in each sampling frequency group as shown in S107. At this time, if decoding is not normally performed, decoding is attempted again using the next sampling frequency value. By repeating this, if the decoding is normally performed, the process of checking whether the current playing speed is normal to the user while playing is checked (S108). If it is faster than the actual playback speed, playback is performed by switching to a lower sampling frequency value in the sampling frequency group as shown in S109.

8 is a diagram illustrating an example of a program to which the audio stream decoding method described above is applied.

As shown in FIG. 8, when the input AAC file is a low stream having a sampling frequency of 16khz, one sampling frequency group is applied one by one until the portion to be properly decoded from the first 48khz (48khz, 32khz, 24khz, 22.050). Khz, 16khz) is attempted, and 16khz is correctly decoded and selected and reproduced as a sampling frequency.

In the present invention, the file format is analyzed for a file whose header does not appear at the beginning of the file. When the analyzed file format is a low stream, the sampling frequency can be found, decoded, and played back while the sampling frequency is sequentially changed according to the sampling frequency group. By doing so, there is an effect that can flexibly respond to the error content to the AAC player received by wire and wireless.

Claims (8)

In the AAC-based audio data decoding method, Searching for the audio stream; Checking the presence or absence of the stream header; Performing decoding in a corresponding format when the stream header is found; During the decoding, if the decoding is not normally performed, repeating the decoding process at a start position of the next frame a predetermined number of times, and recognizing it as a stream if the decoding is not normally performed; And decoding when the raw stream is recognized using the sampling frequency information. The method of claim 1, And the stream header is an ADIF or ADTS header. The method of claim 1, And if the stream header is not found, recognize the stream as a raw stream. The method of claim 1, The decoding step for the raw stream, Finding a sampling frequency candidate group normally decoded for the previous frame; And reproducing at a high sampling frequency among the candidate groups. The method of claim 4, wherein When reproducing at the normal speed when reproducing at the sampling frequency, and reproducing at a low sampling frequency from the candidate group until reproducing at the normal speed if the reproducing is not performed at the normal speed. A decoding control method of an audio frame. The method of claim 5 And determining whether to switch to a low sampling frequency after receiving confirmation from the user of whether to play at the normal speed. In the AAC-based audio data decoding method, Analyzing the format of the file without the ADIF or ADTS header at the beginning of the file; And, in the case of a low stream, finding and reproducing a sampling frequency suitable for a reproduction speed while sequentially changing the sampling frequency according to a sampling frequency group. The method of claim 7, wherein And performing decoding while changing from the highest sampling frequency in the sampling frequency group to a lower sampling frequency step by step.

Images