Stores Channel matrixing method.
Member of Audio Frame
Method 0 (No channel matrixing):
No channel matrixing bits are stored. Channels are stored separately and independently.
- | 0 bit |
Method 1 (Fixed Channel Matrixing after Quantization):
No channel matrixing bits are stored. A fixed matrixing is used AFTER quantization. If this increases bitrate within a frame, the frames method can be switched for that frame to Method 0 or Method 2.
- | 0 bit |
Method 2 (Variable channel matrixing after quantization):
(Active channels - 1) matrixing bits per subband are stored. You can switch between 2 Active channels-1 channel matrixing possiblities which are done after quantization. Therefore there is no channel cross talk, but you can't do channel unmasking compensation for signals with high channel correlation. If this is necessary, switch to Method 3. When signals are highly correlated channel crosstalk is no problem.
Channel matrixing | (Active Channels - 1) bit |
Method 3 (Variable channel matrixing before quantization):
(Active channels - 1) matrixing bits per subband are stored. You can switch between 2 Active channels-1 channel matrixing possiblities which are done before quantization. Therefore there is channel cross talk, but you can do channel unmasking compensation for signals with high channel correlation. If channel cross talk disturbs, switch to Method 2.
Channel matrixing | (Active Channels - 1) bit |
Two unsed bits are stored. Value must be zero for both bits.
Method 0 (No channel matrixing):
Global Channel Matrixing Bits | Mode |
0 | valid |
1 . . . 3 | invalid |
Channel | Contents |
0 | C |
Depending on Global channel matrixing:
L = quantized left channel, R = quantized right channel, l = left channel, r = right channel
Method 0 (No channel matrixing):
Channel | Contents |
0 | L |
1 | R |
Method 1 (Fixed Channel Matrixing after Quantization):
Channel | Contents |
0 | (L + R) >> 1 |
1 | L - R |
Method 2 (Variable channel matrixing after quantization):
For every used subband a MS-Bit is read. Matrixing depends on this bit:
Channel | Contents if MS-Bit = 0 | Contents if MS-Bit = 1 |
0 | L | (L + R) >> 1 |
1 | R | L - R |
Method 3 (Variable channel matrixing before quantization):
For every used subband a MS-Bit is read. Matrixing depends on this bit:
Channel | Contents if MS-Bit = 0 | Contents if MS-Bit = 1 |
0 | l | 0.7071 * (l + r) |
1 | r | 0.7071 * (l - r) |
Recommendations:
Method | Recommended for . . . |
0 | Fast encoding, typical stereo of the 60's (left: voice, right: instruments) |
1 | Fast encoding, Dual stereo with high correlation (same soundtrack, only different voices) |
2 | Normal encoding (no Xtalk, but problems with stereo unmasking), Dual stereo |
3 | Normal encoding (Xtalk, no problems with stereo unmasking) |
Method 0 (No channel matrixing):
Channel | Contents |
0 | L |
1 | C |
2 | R |
Method 1 (Fixed Channel Matrixing after Quantization):
currently undefined.
Method 2 (Variable channel matrixing after quantization):
currently undefined.
Method 3 (Variable channel matrixing before quantization):
For every used subband 2 MS-Bits are read. Matrixing depends on these bits:
Channel | Contents if MS-Bits = 0 | Contents if MS-Bits = 1 | Contents if MS-Bits = 2 | Contents if MS-Bits = 3 |
0 | l | 0.5 * (l + c) | l | 0.5 * (l + r) |
1 | c | 0.5 * (l - c) | 0.5 * (c + r) | c |
2 | r | r | 0.5 * (c - r) | 0.5 * (l - r) |