fenix
/
fdk-aac
mirror of https://github.com/mstorsjo/fdk-aac.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Snap for 6206568 from 925092e8fc to rvc-release 

Change-Id: Ide088e86f1ced06e7717f8d78d187058bc7db370
This commit is contained in:
android-build-team Robot 2020-02-14 03:28:18 +00:00
parent 4e14250c10 925092e8fc
commit 67aa964d42
4 changed files with 53 additions and 104 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

BIN
documentation/aacDecoder.pdf
View File

Binary file not shown.

BIN
documentation/aacEncoder.pdf
View File

Binary file not shown.

116
libAACdec/include/aacdecoder_lib.h
View File

@ -164,9 +164,6 @@ The contents of each file is described in detail in this document. All header
files are provided for usage in specific C/C++ programs. The main AAC decoder files are provided for usage in specific C/C++ programs. The main AAC decoder
library API functions are located in aacdecoder_lib.h header file. library API functions are located in aacdecoder_lib.h header file.
In binary releases the decoder core resides in statically linkable libraries,
for example libAACdec.a.
\section Calling_Sequence Calling Sequence \section Calling_Sequence Calling Sequence
@ -174,19 +171,7 @@ The following sequence is necessary for proper decoding of ISO/MPEG-2/4 AAC,
HE-AAC v2, or MPEG-D USAC bitstreams. In the following description, input stream HE-AAC v2, or MPEG-D USAC bitstreams. In the following description, input stream
read and output write function details are left out, since they may be read and output write function details are left out, since they may be
implemented in a variety of configurations depending on the user's specific implemented in a variety of configurations depending on the user's specific
requirements. The example implementation uses file-based input/output, and in requirements.
such case one may call mpegFileRead_Open() to open an input file and to allocate
memory for the required structures, and the corresponding mpegFileRead_Close()
to close opened files and to de-allocate associated structures.
mpegFileRead_Open() will attempt to detect the bitstream format and in case of
MPEG-4 file format or Raw Packets file format (a proprietary Fraunhofer IIS file
format suitable only for testing) it will read the Audio Specific Config data
(ASC). An unsuccessful attempt to recognize the bitstream format requires the
user to provide this information manually. For any other bitstream formats that
are usually applicable in streaming applications, the decoder itself will try to
synchronize and parse the given bitstream fragment using the FDK transport
library. Hence, for streaming applications (without file access) this step is
not necessary.
-# Call aacDecoder_Open() to open and retrieve a handle to a new AAC decoder -# Call aacDecoder_Open() to open and retrieve a handle to a new AAC decoder
@ -205,19 +190,17 @@ do {
working memory (a client-supplied input buffer "inBuffer" in framework). This working memory (a client-supplied input buffer "inBuffer" in framework). This
buffer will be used to load AAC bitstream data to the decoder. Only when all buffer will be used to load AAC bitstream data to the decoder. Only when all
data in this buffer has been processed will the decoder signal an empty buffer. data in this buffer has been processed will the decoder signal an empty buffer.
For file-based input, you may invoke mpegFileRead_Read() to acquire new
bitstream data.
-# Call aacDecoder_Fill() to fill the decoder's internal bitstream input buffer -# Call aacDecoder_Fill() to fill the decoder's internal bitstream input buffer
with the client-supplied bitstream input buffer. Note, if the data loaded in to with the client-supplied bitstream input buffer. Note, if the data loaded in to
the internal buffer is not sufficient to decode a frame, the internal buffer is not sufficient to decode a frame,
aacDecoder_DecodeFrame() will return ::AAC_DEC_NOT_ENOUGH_BITS until a aacDecoder_DecodeFrame() will return ::AAC_DEC_NOT_ENOUGH_BITS until a
sufficient amount of data is loaded in to the internal buffer. For streaming sufficient amount of data is loaded in to the internal buffer. For streaming
formats (ADTS, LOAS), it is acceptable to load more than one frame to the formats (ADTS, LOAS), it is acceptable to load more than one frame to the
decoder. However, for RAW file format (Fraunhofer IIS proprietary format), only decoder. However, for packed based formats, only one frame may be loaded to the
one frame may be loaded to the decoder per aacDecoder_DecodeFrame() call. For decoder per aacDecoder_DecodeFrame() call. For least amount of communication
least amount of communication delay, fill and decode should be performed on a delay, fill and decode should be performed on a frame by frame basis. \code
frame by frame basis. \code ErrorStatus = aacDecoder_Fill(aacDecoderInfo, ErrorStatus = aacDecoder_Fill(aacDecoderInfo, inBuffer, bytesRead,
inBuffer, bytesRead, bytesValid); \endcode bytesValid); \endcode
-# Call aacDecoder_DecodeFrame(). This function decodes one frame and writes -# Call aacDecoder_DecodeFrame(). This function decodes one frame and writes
decoded PCM audio data to a client-supplied buffer. It is the client's decoded PCM audio data to a client-supplied buffer. It is the client's
responsibility to allocate a buffer which is large enough to hold the decoded responsibility to allocate a buffer which is large enough to hold the decoded
@ -225,12 +208,9 @@ output data. \code ErrorStatus = aacDecoder_DecodeFrame(aacDecoderInfo,
TimeData, OUT_BUF_SIZE, flags); \endcode If the bitstream configuration (number TimeData, OUT_BUF_SIZE, flags); \endcode If the bitstream configuration (number
of channels, sample rate, frame size) is not known a priori, you may call of channels, sample rate, frame size) is not known a priori, you may call
aacDecoder_GetStreamInfo() to retrieve a structure that contains this aacDecoder_GetStreamInfo() to retrieve a structure that contains this
information. You may use this data to initialize an audio output device. In the information. You may use this data to initialize an audio output device. \code
example program, if the number of channels or the sample rate has changed since p_si = aacDecoder_GetStreamInfo(aacDecoderInfo);
program start or the previously decoded frame, the audio output device is then \endcode
re-initialized. If WAVE file output is chosen, a new WAVE file for each new
stream configuration is be created. \code p_si =
aacDecoder_GetStreamInfo(aacDecoderInfo); \endcode
-# Repeat steps 5 to 7 until no data is available to decode any more, or in case -# Repeat steps 5 to 7 until no data is available to decode any more, or in case
of error. \code } while (bytesRead[0] > 0 || doFlush || doBsFlush || of error. \code } while (bytesRead[0] > 0 || doFlush || doBsFlush ||
forceContinue); \endcode forceContinue); \endcode
@ -239,7 +219,7 @@ structures. \code aacDecoder_Close(aacDecoderInfo); \endcode
\image latex decode.png "Decode calling sequence" width=11cm \image latex decode.png "Decode calling sequence" width=11cm
\image latex change_source.png "Change data source sequence" width 5cm \image latex change_source.png "Change data source sequence" width=5cm
\image latex conceal.png "Error concealment sequence" width=14cm \image latex conceal.png "Error concealment sequence" width=14cm
@ -296,16 +276,14 @@ input buffer, and one to hold the decoded output PCM sample data. In resource
limited applications, the output buffer may be reused as an external input limited applications, the output buffer may be reused as an external input
buffer prior to the subsequence aacDecoder_Fill() function call. buffer prior to the subsequence aacDecoder_Fill() function call.
The external input buffer is set in the example program and its size is defined To feed the data to the decoder-internal input buffer, use the
by ::IN_BUF_SIZE. You may freely choose different buffer sizes. To feed the data function aacDecoder_Fill(). This function returns important information
to the decoder-internal input buffer, use the function aacDecoder_Fill(). This regarding the number of bytes in the external input buffer that have not yet
function returns important information regarding the number of bytes in the been copied into the internal input buffer (variable bytesValid). Once the
external input buffer that have not yet been copied into the internal input external buffer has been fully copied, it can be completely re-filled again. In
buffer (variable bytesValid). Once the external buffer has been fully copied, it case you wish to refill the buffer while there are unprocessed bytes (bytesValid
can be completely re-filled again. In case you wish to refill the buffer while is unequal 0), you should preserve the unconsumed data. However, we recommend to
there are unprocessed bytes (bytesValid is unequal 0), you should preserve the refill the buffer only when bytesValid returns 0.
unconsumed data. However, we recommend to refill the buffer only when bytesValid
returns 0.
The bytesValid parameter is an input and output parameter to the FDK decoder. As The bytesValid parameter is an input and output parameter to the FDK decoder. As
an input, it signals how many valid bytes are available in the external buffer. an input, it signals how many valid bytes are available in the external buffer.
@ -340,10 +318,7 @@ explanation, please refer to ISO/IEC 13818-7:2005(E), chapter 8.5.3.2.
In case a Program Config is included in the audio configuration, the channel In case a Program Config is included in the audio configuration, the channel
mapping described within it will be adopted. mapping described within it will be adopted.
In case of MPEG-D Surround the channel mapping will follow the same criteria The examples below explain these aspects in detail.
described in ISO/IEC 13818-7:2005(E), but adding corresponding top channels (if
available) to the channel types in order to avoid ambiguity. The examples below
explain these aspects in detail.
\section OutputFormatChange Changing the audio output format \section OutputFormatChange Changing the audio output format
@ -689,9 +664,7 @@ typedef enum {
2. If the parameter value is greater than that of 2. If the parameter value is greater than that of
::AAC_PCM_MAX_OUTPUT_CHANNELS both will be set to the same ::AAC_PCM_MAX_OUTPUT_CHANNELS both will be set to the same
value. \n value. \n
3. This parameter does not affect MPEG Surround processing. 3. This parameter will be ignored if the number of encoded
\n
4. This parameter will be ignored if the number of encoded
audio channels is greater than 8. */ audio channels is greater than 8. */
AAC_PCM_MAX_OUTPUT_CHANNELS = AAC_PCM_MAX_OUTPUT_CHANNELS =
0x0012, /*!< Maximum number of PCM output channels. If lower than the 0x0012, /*!< Maximum number of PCM output channels. If lower than the
@ -718,11 +691,7 @@ typedef enum {
2. If the parameter value is greater than zero but smaller 2. If the parameter value is greater than zero but smaller
than ::AAC_PCM_MIN_OUTPUT_CHANNELS both will be set to same than ::AAC_PCM_MIN_OUTPUT_CHANNELS both will be set to same
value. \n value. \n
3. The operating mode of the MPEG Surround module will be 3. This parameter will be ignored if the number of encoded
set accordingly. \n
4. Setting this parameter with any value will disable the
binaural processing of the MPEG Surround module
5. This parameter will be ignored if the number of encoded
audio channels is greater than 8. */ audio channels is greater than 8. */
AAC_METADATA_PROFILE = AAC_METADATA_PROFILE =
0x0020, /*!< See ::AAC_MD_PROFILE for all available values. */ 0x0020, /*!< See ::AAC_MD_PROFILE for all available values. */
@ -803,11 +772,11 @@ typedef enum {
sequences for fading in and out, if provided in the sequences for fading in and out, if provided in the
bitstream.\n Enabled album mode makes use of dedicated album bitstream.\n Enabled album mode makes use of dedicated album
loudness information, if provided in the bitstream.\n */ loudness information, if provided in the bitstream.\n */
AAC_QMF_LOWPOWER = 0x0300, /*!< Quadrature Mirror Filter (QMF) Bank processing AAC_QMF_LOWPOWER =
mode. \n -1: Use internal default. Implies MPEG 0x0300, /*!< Quadrature Mirror Filter (QMF) Bank processing mode. \n
Surround partially complex accordingly. \n 0: -1: Use internal default. \n
Use complex QMF data mode. \n 1: Use real (low 0: Use complex QMF data mode. \n
power) QMF data mode. \n */ 1: Use real (low power) QMF data mode. \n */
AAC_TPDEC_CLEAR_BUFFER = AAC_TPDEC_CLEAR_BUFFER =
0x0603 /*!< Clear internal bit stream buffer of transport layers. The 0x0603 /*!< Clear internal bit stream buffer of transport layers. The
decoder will start decoding at new data passed after this event decoder will start decoding at new data passed after this event
@ -1038,21 +1007,24 @@ LINKSPEC_H AAC_DECODER_ERROR aacDecoder_Fill(HANDLE_AACDECODER self,
const UINT bufferSize[], const UINT bufferSize[],
UINT *bytesValid); UINT *bytesValid);
#define AACDEC_CONCEAL \ /** Flag for aacDecoder_DecodeFrame(): Trigger the built-in error concealment
1 /*!< Flag for aacDecoder_DecodeFrame(): Trigger the built-in error \ * module to generate a substitute signal for one lost frame. New input data
concealment module to generate a substitute signal for one lost frame. \ * will not be considered.
New input data will not be considered. */ */
#define AACDEC_FLUSH \ #define AACDEC_CONCEAL 1
2 /*!< Flag for aacDecoder_DecodeFrame(): Flush all filterbanks to get all \ /** Flag for aacDecoder_DecodeFrame(): Flush all filterbanks to get all delayed
delayed audio without having new input data. Thus new input data will \ * audio without having new input data. Thus new input data will not be
not be considered.*/ * considered.
#define AACDEC_INTR \ */
4 /*!< Flag for aacDecoder_DecodeFrame(): Signal an input bit stream data \ #define AACDEC_FLUSH 2
discontinuity. Resync any internals as necessary. */ /** Flag for aacDecoder_DecodeFrame(): Signal an input bit stream data
#define AACDEC_CLRHIST \ * discontinuity. Resync any internals as necessary.
8 /*!< Flag for aacDecoder_DecodeFrame(): Clear all signal delay lines and \ */
history buffers. CAUTION: This can cause discontinuities in the output \ #define AACDEC_INTR 4
signal. */ /** Flag for aacDecoder_DecodeFrame(): Clear all signal delay lines and history
* buffers. CAUTION: This can cause discontinuities in the output signal.
*/
#define AACDEC_CLRHIST 8
/** /**
* \brief Decode one audio frame * \brief Decode one audio frame

41
libAACenc/include/aacenc_lib.h
View File

@ -1,7 +1,7 @@
/* ----------------------------------------------------------------------------- /* -----------------------------------------------------------------------------
Software License for The Fraunhofer FDK AAC Codec Library for Android Software License for The Fraunhofer FDK AAC Codec Library for Android
© Copyright 1995 - 2018 Fraunhofer-Gesellschaft zur Förderung der angewandten © Copyright 1995 - 2019 Fraunhofer-Gesellschaft zur Förderung der angewandten
Forschung e.V. All rights reserved. Forschung e.V. All rights reserved.
1. INTRODUCTION 1. INTRODUCTION
@ -149,12 +149,6 @@ All API header files are located in the folder /include of the release package.
All header files are provided for usage in C/C++ programs. The AAC encoder All header files are provided for usage in C/C++ programs. The AAC encoder
library API functions are located in aacenc_lib.h. library API functions are located in aacenc_lib.h.
In binary releases the encoder core resides in statically linkable libraries
called for example libAACenc.a/libFDK.a (LINUX) or FDK_fastaaclib.lib (MS Visual
C++) for the plain AAC-LC core encoder and libSBRenc.a (LINUX) or
FDK_sbrEncLib.lib (MS Visual C++) for the SBR (Spectral Band Replication) and PS
(Parametric Stereo) modules.
\section CallingSequence Calling Sequence \section CallingSequence Calling Sequence
For encoding of ISO/MPEG-2/4 AAC bitstreams the following sequence is mandatory. For encoding of ISO/MPEG-2/4 AAC bitstreams the following sequence is mandatory.
@ -326,18 +320,12 @@ input buffer, simulating a modulo buffer: \code if (outargs.numInSamples>0) {
\endcode \endcode
\section writeOutData Output Bitstream Data \section writeOutData Output Bitstream Data
If any AAC bitstream data is available, write it to output file or device. This If any AAC bitstream data is available, write it to output file or device as
can be done once the following condition is true: \code if follows. \code if (outargs.numOutBytes>0) { FDKfwrite(outputBuffer,
(outargs.numOutBytes>0) { outargs.numOutBytes, 1, pOutFile);
} }
\endcode \endcode
If you use file I/O then for example call mpegFileWrite_Write() from the library
libMpegFileWrite \code mpegFileWrite_Write(hMpegFile, outputBuffer,
outargs.numOutBytes, aacEncoder_GetParam(hAacEncoder, AACENC_GRANULE_LENGTH));
\endcode
\section cfgMetaData Meta Data Configuration \section cfgMetaData Meta Data Configuration
If the present library is configured with Metadata support, it is possible to If the present library is configured with Metadata support, it is possible to
@ -427,7 +415,7 @@ switch (nChannels) {
return chMode; return chMode;
\endcode \endcode
\subsection bitreservoir Bitreservoir Configuration \subsection peakbitrate Peak Bitrate Configuration
In AAC, the default bitreservoir configuration depends on the chosen bitrate per In AAC, the default bitreservoir configuration depends on the chosen bitrate per
frame and the number of effective channels. The size can be determined as below. frame and the number of effective channels. The size can be determined as below.
\f[ \f[
@ -436,17 +424,10 @@ bitreservoir = nEffChannels*6144 - (bitrate*framelength/samplerate)
Due to audio quality concerns it is not recommended to change the bitreservoir Due to audio quality concerns it is not recommended to change the bitreservoir
size to a lower value than the default setting! However, for minimizing the size to a lower value than the default setting! However, for minimizing the
delay for streaming applications or for achieving a constant size of the delay for streaming applications or for achieving a constant size of the
bitstream packages in each frame, it may be necessaray to change the bitstream packages in each frame, it may be necessaray to limit the maximum bits
bitreservoir size. This can be done with the ::AACENC_PEAK_BITRATE parameter. per frame size. This can be done with the ::AACENC_PEAK_BITRATE parameter. \code
\code
aacEncoder_SetParam(hAacEncoder, AACENC_PEAK_BITRATE, value); aacEncoder_SetParam(hAacEncoder, AACENC_PEAK_BITRATE, value);
\endcode \endcode
By setting ::AACENC_BITRATEMODE to fixed framing, the bitreservoir is disabled.
A disabled bitreservoir results in a constant size for each bitstream package.
Please note that especially at lower bitrates a disabled bitreservoir can
downgrade the audio quality considerably! The default bitreservoir configuration
can be achieved as follows. \code aacEncoder_SetParam(hAacEncoder,
AACENC_BITRESERVOIR, -1); \endcode
To achieve acceptable audio quality with a reduced bitreservoir size setting at To achieve acceptable audio quality with a reduced bitreservoir size setting at
least 1000 bits per audio channel is recommended. For a multichannel audio file least 1000 bits per audio channel is recommended. For a multichannel audio file
@ -967,9 +948,7 @@ in this Fraunhofer IIS AAC encoder. AAC has been designed in that way.
\subsection BEHAVIOUR_ESTIM_AVG_FRAMESIZES Estimating Average Frame Sizes \subsection BEHAVIOUR_ESTIM_AVG_FRAMESIZES Estimating Average Frame Sizes
A HE-AAC v1 or v2 audio frame contains 2048 PCM samples per channel (there is A HE-AAC v1 or v2 audio frame contains 2048 PCM samples per channel.
also one mode with 1920 samples per channel but this is only for special
purposes such as DAB+ digital radio).
The number of HE-AAC frames \f$N\_FRAMES\f$ per second at 44.1 kHz is: The number of HE-AAC frames \f$N\_FRAMES\f$ per second at 44.1 kHz is:
@ -1082,9 +1061,7 @@ typedef struct AACENCODER *HANDLE_AACENCODER;
typedef struct { typedef struct {
UINT maxOutBufBytes; /*!< Maximum number of encoder bitstream bytes within one UINT maxOutBufBytes; /*!< Maximum number of encoder bitstream bytes within one
frame. Size depends on maximum number of supported frame. Size depends on maximum number of supported
channels in encoder instance. For superframing (as channels in encoder instance. */
used for example in DAB+), size has to be a multiple
accordingly. */
UINT maxAncBytes; /*!< Maximum number of ancillary data bytes which can be UINT maxAncBytes; /*!< Maximum number of ancillary data bytes which can be
inserted into bitstream within one frame. */ inserted into bitstream within one frame. */