ADTS全稱是(Audio Data Transport Stream),是AAC的一種十分常見的傳輸格式。
記得第一次做demux的時候,把AAC音頻的ES流從FLV封裝格式中抽出來送給硬件解碼器時,不能播;保存到本地用pc的播放器播時,我靠也不能播。當時崩潰了,后來通過查找資料才知道。一般的AAC解碼器都需要把AAC的ES流打包成ADTS的格式,一般是在AAC ES流前添加7個字節的ADTS header。也就是說你可以吧ADTS這個頭看作是AAC的frameheader。
ADTS AAC | ||||||
ADTS_header | AAC ES | ADTS_header | AAC ES | ... | ADTS_header | AAC ES |
ADTS 頭中相對有用的信息 采樣率、聲道數、幀長度。想想也是,我要是解碼器的話,你給我一堆得AAC音頻ES流我也解不出來。每一個帶ADTS頭信息的AAC流會清晰的告送解碼器他需要的這些信息。
一般情況下ADTS的頭信息都是7個字節,分為2部分:
adts_fixed_header();
adts_variable_header();
syncWord :同步頭 總是0xFFF, all bits must be 1,代表著一個ADTS幀的開始
ID:MPEG Version: 0 for MPEG-4, 1 for MPEG-2
Layer:always: '00'
PRofile:表示使用哪個級別的AAC,有些芯片只支持AAC LC 。在MPEG-2 AAC中定義了3種:
sampling_frequency_index:表示使用的采樣率下標,通過這個下標在 Sampling Frequencies[ ]數組中查找得知采樣率的值。
There are 13 supported frequencies:
0: 96000 Hz1: 88200 Hz2: 64000 Hz3: 48000 Hz4: 44100 Hz5: 32000 Hz6: 24000 Hz7: 22050 Hz8: 16000 Hz9: 12000 Hz10: 11025 Hz11: 8000 Hz12: 7350 Hz13: Reserved14: Reserved15: frequency is written explictlychannel_configuration: 表示聲道數 0: Defined in AOT Specifc Config1: 1 channel: front-center2: 2 channels: front-left, front-right3: 3 channels: front-center, front-left, front-right4: 4 channels: front-center, front-left, front-right, back-center5: 5 channels: front-center, front-left, front-right, back-left, back-right6: 6 channels: front-center, front-left, front-right, back-left, back-right, LFE-channel7: 8 channels: front-center, front-left, front-right, side-left, side-right, back-left, back-right, LFE-channel8-15: Reservedframe_length : 一個ADTS幀的長度包括ADTS頭和AAC原始流.
adts_buffer_fullness:0x7FF 說明是碼率可變的碼流
如果是通過嵌入式高清解碼芯片做產品的話,一般情況的解碼工作都是由硬件來完成的。所以大部分的工作是把AAC原始流打包成ADTS的格式,然后丟給硬件就行了。
通過對ADTS格式的了解,很容易就能把AAC打包成ADTS。我們只需得到封裝格式里面關于音頻采樣率、聲道數、元數據長度、aac格式類型等信息。然后在每個AAC原始流前面加上個ADTS頭就OK了。
#define ADTS_HEADER_SIZE 7static int m_sampleRateIndex = 3; // 48.0 KHZstatic int m_channel = 2; // 雙聲道static int m_profile = 1; // AAC(Version 4) LCstatic void add_adts_header(unsigned char *p, int es_len) { int frame_len = ADTS_HEADER_SIZE + es_len; *p++ = 0xff; //syncword (0xfff, high_8bits) *p = 0xf0; //syncword (0xfff, low_4bits) *p |= (0 << 3); //ID (0, 1bit) *p |= (0 << 1); //layer (0, 2bits) *p |= 1; //protection_absent (1, 1bit) p++; *p = (unsigned char) ((m_profile & 0x3) << 6); //profile (profile, 2bits) *p |= ((m_sampleRateIndex & 0xf) << 2); //sampling_frequency_index (sam_idx, 4bits) *p |= (0 << 1); //private_bit (0, 1bit) *p |= ((m_channel & 0x4) >> 2); //channel_configuration (channel, high_1bit) p++; *p = ((m_channel & 0x3) << 6); //channel_configuration (channel, low_2bits) *p |= (0 << 5); //original/copy (0, 1bit) *p |= (0 << 4); //home (0, 1bit); *p |= (0 << 3); //copyright_identification_bit (0, 1bit) *p |= (0 << 2); //copyright_identification_start (0, 1bit) *p |= ((frame_len & 0x1800) >> 11); //frame_length (value, high_2bits) p++; *p++ = (unsigned char) ((frame_len & 0x7f8) >> 3); //frame_length (value, middle_8bits) *p = (unsigned char) ((frame_len & 0x7) << 5); //frame_length (value, low_3bits) *p |= 0x1f; //adts_buffer_fullness (0x7ff, high_5bits) p++; *p = 0xfc; //adts_buffer_fullness (0x7ff, low_6bits) *p |= 0; //number_of_raw_data_blocks_in_frame (0, 2bits); p++;}
新聞熱點
疑難解答