MRC01
Major Contributor
In summary, as I understand the HDCD spec, it uses the LSBs of each sample as a bitstream to encode flags telling the DAC to trigger various functions that the HDCD spec requires to be programmed into the DAC, like different AA filters or amplitude shaping curves.
I'll ignore the AA filter flags because they should be essentially inaudible and even if you did decode them, you might want to ignore them and use the standard linear phase "sharp" filter all the time anyway.
The amplitude shaping functions are more interesting. For example, consider an extended quiet section of music. If it's 36 dB below full scale, the top 6 bits are all zero so they only have 10 bits of resolution (even less in the treble, due to the spectrum of music). In post-processing, they can digitally encode this quiet section at a higher level, getting more bits per sample. If you play it back without HDCD encoding, the quiet part is simply louder so it's like dynamic range compression. As the quiet part arrives, HDCD encodes a flag that tells the DAC to reduce the amplitude. The DAC is working in 24 bit so it can shift it down without loss of resolution. Then as the music gets louder again, the HDCD encodes another flag telling the DAC to "go back to normal".
Several of the CDs in my collection happen to be HDCD encoded. I've decoded them with various PC software and found that none of them actually use this feature. If this is common, then HDCD seems like a marketing scheme to get companies building DACs to pay license fees, and light up the HDCD icon in the DAC display to make consumers feel happy. Or, is this amplitude shaping feature commonly used, and I just have a poor sample of CDs? Or maybe my decoding software is missing the flags?
I'll ignore the AA filter flags because they should be essentially inaudible and even if you did decode them, you might want to ignore them and use the standard linear phase "sharp" filter all the time anyway.
The amplitude shaping functions are more interesting. For example, consider an extended quiet section of music. If it's 36 dB below full scale, the top 6 bits are all zero so they only have 10 bits of resolution (even less in the treble, due to the spectrum of music). In post-processing, they can digitally encode this quiet section at a higher level, getting more bits per sample. If you play it back without HDCD encoding, the quiet part is simply louder so it's like dynamic range compression. As the quiet part arrives, HDCD encodes a flag that tells the DAC to reduce the amplitude. The DAC is working in 24 bit so it can shift it down without loss of resolution. Then as the music gets louder again, the HDCD encodes another flag telling the DAC to "go back to normal".
Several of the CDs in my collection happen to be HDCD encoded. I've decoded them with various PC software and found that none of them actually use this feature. If this is common, then HDCD seems like a marketing scheme to get companies building DACs to pay license fees, and light up the HDCD icon in the DAC display to make consumers feel happy. Or, is this amplitude shaping feature commonly used, and I just have a poor sample of CDs? Or maybe my decoding software is missing the flags?