Ripping HDCD audio

Meaning home grown dither/truncate beats the existing 16 bit presentation (un-decoded) on the disc? Is there a authoritative reference for this?

 

Arguably, yes.

The point is that the encoded HDCD signal is dynamically compressed if PE and/or LLRE are used. What's on CD isn't simply 16-bit, it's compressed 16-bit. When you decode, you're reversing that compression. Basically, the idea behind HDCD is that it's meant to be decoded, but if it's not, it won't sound that different/bad. But it's still not the sound as originally intended.

 

So there is no actual hard data for this, correct? Again, I totally agree in the context of keeping the resulting 24 bit file that hdcd.exe creates. But "I can dither better than Pacific Microsonics" is a rather dubious claim without any hard data IMHO.

 

Hard data for what?

This isn't a matter of one dither vs another. It's restoring the dynamics originally present.

 

Green Day's "Nimrod" is an HDCD disc makes big use of peak extend. The undecoded CD has a DR6. Decoded it has a DR10. It's a very noticeable difference. Converting the decoded version back to 16-bits is going to sound much better than the 16-bit undecoded version.

 

Hypothetical:

Assuming you had a reasonable 24/96 ADC, could you rip/record via analog-out of an HDCD DAC into ripping software, and expect to get reasonable HDCD benefits, even if it's not necessarily bit-perfect?

 

Ah, the consistent voice of reason.

Thanks for this. I concur.

 

That's what we were saying from the start.

 

Sorry. For me, hard data trumps testimony every time.

 

Testimony? What are you talking about? How is describing exactly how the process works not "hard data"?

 

You misunderstood what we were saying.

 

Sure, but you are better off SQ-wise to decode it digitally.

 

As I suspect.

But I'm sure not better off effort-wise. This is a Mac household.

 

The extra dynamic range is the whole point of what I'm doing. And as they are coming out at a lesser volume and more dynamic (to my ears and as measured by TT DR Meter), I think I'm doing it right and getting what I wanted - the less compressed sound of a deconded HDCD track but in a regular 16 bit/44.1 khz file.

For discs that are HDCD decoded without peak extend (like the older Audio Fidelity Gold CD's) there would be no point in doing the conversion.

 

For car/home RBCD listening I decoded several PE/LLRE-encoded HDCDs (i.e. Joni Mitchell's Hits & Misses, Green Day's Nimrod, etc.) to padded 24/44 form, normalized compilation tracks to equal perceived loudness based on track RG, added some pre-amplification/attenuation (where needed), downrezzed them to CDDA with non-intrusive NS dither (mda VST plug-in) in fb2k & then burnt me some CD-TEXT enabled CD-ROMs. Worked great! The sound quality of these home-made discs exceeds that of non-decoded HDCDs, IMO...

 

Is it worth using the plug-in and decoding non-peak extend CDs to the padded 24bit? Since about 2012 or so, the Grateful Dead releases have not had peak extend enabled, and it seems like the other "HDCD" discs in my collection are hit or miss with PE.

 

To the best of my knowledge, no, not worth it if the disc doesn't have peak extension.

 

*Or Low Level Range Extend.

 

Thanks!

 

THIS.

And if you use a good dither, like MBit+ to bring that 24-bit decoded file back down to 16-bit, the resultant file is still going to sound pret-ty, pret-ty, pret-ty, pret-ty good. IMO.

 

This is all fascinating stuff (no sarcasm), but a question I asked before: Out of a OPPO 103 with HDCD enabled, what will be the digital output from the coax/optical? Oppo says the digital output will be 24 bits out of the HDMI, but I find it counterintuitive that the coax/optical stream would be different (not that there is a technical reason for it to not be. It can be whatever it was the designer wanted it to be). Can anybody get an actual read on the bit depth of the coax or the optical output when playing a HDCD?

 

I don't know offhand, but I could test it if nobody else happens to know already.

 

Please do. I don't have HDMI into my amp, just the tv. Would be very grateful if you could put this question to rest for me.

I think I know the answer, by listening, but I would like to have hard data, not my biased/placebo'ed opinion.

I agree there are HDCDs where there is no difference really, but in others, the difference is like, well, high definition audio!

 

I cannot give you precisely the answer you are looking for, but maybe it will be somewhat helpful if I pass on what I've recently learned...

I have an Oppo BDP-93. I thought I'd see how it did decoding HDCD for use in a set of album FLACs. I used my HDCD of Joni Mitchell's "For The Roses", and ran the Oppo's coax out into my Sound Devices USBPre 2 (a real "Swiss army knife" of ADC and DAC audio connections). I ran the USB out of the USBPre 2 into a laptop hosting Sony Vegas Pro 12. While I had no direct way to read sampling bit depth, I set Vegas Pro to record 24/44.1k from the laptop's USB in. After recording "For the Roses" from the Oppo by this method, I compared the peak and RMS results with the peak and RMS levels of a conventional (non-HDCD) CD rip of the same Joni Mitchell disc. While I did not retain those exact figures, I can confirm that for the specific tracks I compared, for each track the "peak - RMS" dB level of the Oppo's coax out was clearly a numerically greater value than the "peak - RMS" dB level of the straight CD rip of those same tracks. Also, the coax recorded tracks were "quieter" (reduced RMS levels) compared to the straight CD rip levels.

A few days later I conducted the same effort with Patty Hurst Shifter's "Too Crowded On The Losing End". Both Windows Media Player and the Oppo 93 indicated that this album too had the HDCD flag set. However, with the Patty Hurst Shifter album I think the HDCD flag was falsely set because the Oppo coax/USBPre 2/Vegas Pro recorded version and the straight CD rip were identical to each other in all respects. In fact, with the "Shifter" album, the coax recorded RMS levels were absolutely identical to the rip RMS levels.

Since I set recording bit depth manually to 24 bits, there may very well have been eight bits of low-level padding added in Vegas Pro. However with "For The Roses", both my numerical comparisons and actual listening confirmed that the Oppo BDP-93 coax out HDCD recording provided clearly better quality than a conventional CD rip of the same album.

 

Not all HDCD discs make use of Peak Extend. Peak Extend is the HDCD feature that uncompresses peaks and can result in the decoded audio having a measurably higher RMS or DR value. If an HDCD title does not make use of Peak Extend then there is not likely to be a measurable difference in the RMS or DR value for the decoded vs. undecoded tracks.

Here's a post where I list the HDCD features present in the HDCD titles that I have (well, the titles that I had back in 2012). Find titles that you have in that list that makes use of Peak Extend. Compare those. The thread and post: HDCDs With Peak Extend

If you want an easy title to test for an audible difference between decoded vs. undecoded then try Green Day "Nimrod". That album makes use of Peak Extend. And makes use of it in a very noticeable way, and in a way that is more noticeable than any other HDCD title that I've tried.