Luoda development - audio stream processing - AAC / loopbacktest as an example

Audio streaming is mainly in DSP Intermediate processing , Here to aac_dec_interfaces.c and audio_loopback_test_interface.c Two source files to analyze and introduce with examples . The paths of the two files are ：

bta_sdk\dsp\middleware\MTK\dspalg\aac_decoder\src
bta_sdk\dsp\middleware\MTK\dspalg\audio_loopback_test\src

among aac_dec_interface.c There are also in another position ：bta_sdk\dsp\middleware\third_party\dspalg\aac_decoder\src

It can be seen from the path that this is a third party ACC Coding function .

1、aac_dec_interfaces.c[MTK\dspalg\aac_decoder\src]

  The function of the function passes the input audio data through AAC After encoding, it is sent to the left and right output channels ; if AAC The processing module has not been initialized , Then initialize first and then process the data ; The interface function in the file is ：

bool stream_codec_decoder_aac_process (void *para);

The flow of the function is shown in the following figure ：

 

AAC The encoding function is ：

int AIR_AAC_decoder(void *AAC_INSTANCE, uint8_t *input_buf, int aac_packet_len , int32_t *output_L, int32_t *output_R, int *decoded_len);

Parameters AAC_INSTANCE Used to store AAC Instance object of function module initialization , Its value is due to AAC The first parameter of the initialization function returns ：

int AIR_AAC_init(void *AAC_INSTANCE, uint8_t *input_buf, int aac_packet_len , int16_t *ch, uint32_t *srate, int16_t *brate);

Parameters aac_packet_len by input_buf The length of the audio stream to be processed ;

Parameters input_buf For input audio data , From to Bluetooth channel , For example, the sound stream from the mobile phone ;

Parameters output_L Is the audio stream pointing to the left channel output buffer;

Parameters output_R Is the audio stream output to the right channel buffer;

Parameters decoded_len Express AAC Encode the length of data actually processed ;

From which we can know , The source of the audio stream is 1  Input channels , It flows to No 1 And the 2 A passage , Its interface functions are ：

void* stream_codec_get_1st_input_buffer(void* para)
void* stream_codec_get_1st_output_buffer(void* para)
void* stream_codec_get_2nd_output_buffer(void* para)

The position of the interface function is ：

bta_sdk\dsp\middleware\MTK\dspfw\port\chip\ab156x\src\dsp_lower_layer\dsp_interface\ dsp_feature_interface.c

Analysis code can know , The output channels are 6 individual , Defined by the following macro ：

#define CALLBACK_OUTPUT_PORT_MAX_NUM  6

Input channels share 5 individual , Defined by the following macro ：

#define CALLBACK_INPUT_PORT_MAX_NUM   5

The general function to get the cache address of the specified channel is ：

void* stream_codec_get_input_buffer(void* para, uint32_t channel)
void* stream_codec_get_output_buffer(void* para, uint32_t channel)

The following function , It can be used to obtain the length of the input data stream ：

uint16_t stream_codec_get_input_size(void* para)

In addition, there is the data width of the output channel , be for the purpose of 16 It's still 32 Bit data format ：

uint8_t stream_codec_get_output_resolution(void* para)

These functions are used to obtain the cache first address of the corresponding channel data , Then get the length of the data , In this way, these data can be operated .

2、audio_loopback_test_interface.c[bta_sdk\dsp\middleware\MTK\dspalg\audio_loopback_test\src]

 

 

 

Function call flow ：

 

3、 Definition of input and output buffer

  that , Another question , How to define the storage fetches obtained by these interfaces , That is to say 1 individual inout buffer, The first 1 individual input buffer Point to the buffer Where is it , Who will go to this buffer Put data in it ？

stream_function_get_1st_inout_buffer
stream_function_get_2nd_inout_buffer
void* stream_codec_get_1st_input_buffer(void* para)
void* stream_codec_get_1st_output_buffer(void* para)
void* stream_codec_get_2nd_output_buffer(void* para)

You can know by code , The first address of the storage area is determined by out_ptr[] and in_ptr[] Defined by ：

typedef struct DSP_Entry_Para_s {
    VOID*                   mem_ptr;

    U16                     in_malloc_size;
    U16                     in_size;
    VOID*                   in_ptr[CALLBACK_INPUT_PORT_MAX_NUM];

    U16                     out_malloc_size;
    U16                     codec_out_size;
    VOID*                   out_ptr[CALLBACK_OUTPUT_PORT_MAX_NUM];

    U8                      in_channel_num;
    U8                      in_sampling_rate;
    U8                      out_channel_num;
    U8                      device_out_channel_num;
    U8                      software_handled_channel_num;
    U8                      codec_out_sampling_rate;

    TaskHandle_t            DSPTask;

    U8                      with_encoder;
    U16                     encoder_out_size;

    U8                      with_src;
    U8                      src_out_sampling_rate;
    U16                     src_out_size;

    U16                     pre_codec_out_size;
    U8                      pre_codec_out_sampling_rate;
    U8                      skip_process;
    U8                      bypass_mode;
    bool                    force_resume;
    DSP_FEATURE_RESOLUTION  resolution;
    DSP_FEATURE_NUMBERING   number;
    DSP_PKT_INFO_STRU       pkt_info;
    #ifdef PRELOADER_ENABLE
    VOID*                   feature_ptr;
    #endif
} DSP_ENTRY_PARA, *DSP_ENTRY_PARA_PTR;

These two specified initializations , It's in the file dsp_callback.c->DSP_Callback_ParaSetup(DSP_STREAMING_PARA_PTR) Finish in , As shown in the following code .

 

if (stream->callback.FeatureTablePtr->FeatureType != CODEC_PCM_COPY) {
        /* malloc in_ptr */
        configASSERT(stream->callback.EntryPara.in_channel_num<=CALLBACK_INPUT_PORT_MAX_NUM);
        mallocSize = stream->callback.EntryPara.in_malloc_size*stream->callback.EntryPara.in_channel_num;
        mem_ptr = DSPMEM_tmalloc(stream->callback.EntryPara.DSPTask, mallocSize, stream);
        memset(mem_ptr, 0, mallocSize);
        for (i=0 ; i<stream->callback.EntryPara.in_channel_num ; i++)
        {
            stream->callback.EntryPara.in_ptr[i] = mem_ptr;
            mem_ptr += stream->callback.EntryPara.in_malloc_size;
        }
        /* malloc out_ptr */
        chNum = MAX(stream->callback.EntryPara.out_channel_num, 2);
        configASSERT(chNum<=CALLBACK_OUTPUT_PORT_MAX_NUM);
        mallocSize = stream->callback.EntryPara.out_malloc_size*chNum;
        mem_ptr = DSPMEM_tmalloc(stream->callback.EntryPara.DSPTask, mallocSize, stream);
        memset(mem_ptr, 0, mallocSize);
        for (i=0 ; i<chNum ; i++)
        {
            stream->callback.EntryPara.out_ptr[i]= mem_ptr;
            mem_ptr += stream->callback.EntryPara.out_malloc_size;
        }
     } else {
        //CODEC_PCM_COPY
        chNum = MAX(MAX(stream->callback.EntryPara.out_channel_num, 1), stream->callback.EntryPara.in_channel_num);
        configASSERT(chNum<=MIN(CALLBACK_INPUT_PORT_MAX_NUM, CALLBACK_OUTPUT_PORT_MAX_NUM));
        frameSize = MAX(stream->callback.EntryPara.in_malloc_size, stream->callback.EntryPara.out_malloc_size);
        stream->callback.EntryPara.in_malloc_size = stream->callback.EntryPara.out_malloc_size = frameSize;
        stream->callback.EntryPara.out_channel_num = chNum;

        mallocSize = frameSize*chNum;
        if ((frameSize & 3) && (chNum > 1))
        {
            DSP_MW_LOG_I("[DSP] Unaligned Callback Frame Size:%d!!", 1, frameSize);
        }
        mem_ptr = DSPMEM_tmalloc(stream->callback.EntryPara.DSPTask, mallocSize, stream);
        memset(mem_ptr, 0, mallocSize);
        for (i=0 ; i<chNum ; i++)
        {
            stream->callback.EntryPara.out_ptr[i]= mem_ptr;
            stream->callback.EntryPara.in_ptr[i] = mem_ptr;
            mem_ptr += frameSize;
        }
        DSP_MW_LOG_I("[DSP] Callback stream Setup codec is CODEC_PCM_COPY, Frame Size:%d, channel_num:%d", 2, frameSize, chNum);
    }

function stream_codec_decoder_aac_process(…) In the file dsp_sdk.c Is indexed to data stream_feature_table[] in , As shown in the figure below ：

 

So by analyzing , You can know CODEC_DECODER_AAC This type of , Is in A2DP Is used in , And it is used when decoding , Because you can guess ,AAC Is to serve A2DP Decode the incoming audio , And then through Speaker Play out .

 

So it is not difficult to understand ,

bool stream_codec_decoder_aac_process (void *para)

Function stream_codec_get_1st_input_buffer  It means A2DP Of Source Incoming audio stream , and stream_codec_get_1st_output_buffer

and

stream_codec_get_2nd_output_buffer

It's left and right Speaker Output data stream , Flow direction left and right Speaker Two channels ;

1. about A2DP, in_channel_num=1, The definition is shown in the figure below ：
   
2. about A2DP, out_channel_num=2, As shown in the figure below ：
   

  As indicated by the red line ,out_channel_num by 2 perhaps audio.channel_num（ as for channel_num What's the value of , Not studied yet , Guess it should be 1 perhaps 2）;