How to rip 20-bit CD(and retain all 20bits) Options

[ben]

I have a 20-bit Telarc CD, and would like to extract it onto my computer. However, I'm not really sure how to go about this without having the WAV truncated to 16-bit. I only have a 16bit soundcard right now, so it's going to get chopped off anyway, but I just want the original, unaltered music in the file.

Any ideas?

[bryant]

Audio CDs store all the audio information in only 16 bits. When a CD advertises that it is 20-bit, this simply means that it came from a higher resolution source and that special processing was used to retain as much fidelity as possible when finally converted to the 16-bit resultion of CDs. Usually this means that 20 bits of dynamic range has been preserved, but only at certain frequencies. Even HDCD disc store all their information in the standard 16-bit audio word.

I would suggest that to maintain the fidelity you refrain from doing any signal processing on the audio during playback, like normalization or anything else that involves dither that can essentially erase the higher resolution. If you avoid doing this you will get the full resolution through your 16-bit soundcard.

[Jebus]

I am assuming it is a HDCD. This has been discussed many times here with (AFAIKT) no real conclusions made. Some people insist that 20-bits cannot be extracted without special hardware. Some think it is just marketing, and there is actually no additional bitwidth on the disc at all. It appears to me that the 20-bit playback of HDCDs is a proprietary feature of windows media player 9 (and some hardware players), and therefore no ripping utilities exist.

I really wouldn't worry about it though - the extra 4 bits just give you more dynamic range, or which most modern music doesn't use. (They don't even use the 16-bits of regular redbook, in the interest of compression). Ripping as per normal to 16-bit wave should not (in theory) loose any critical information as a result.

Compression will eventually dither down to 16 bits anyhow, so don't sweat it!

edit: oops, put SACD instead of HDCD. fixed

This post has been edited by Jebus: May 8 2003, 21:01

[ben]

OK cool..

And I just noticed, under the "Technical Information" part of the CD booklet, it says: "20- to 16-bit Encoding: Apogee UV-1000 Super CD Encoder"

Thanks for clearing that up..

This post has been edited by ben: May 8 2003, 03:32

[tigre]

I am assuming it is a SACD. This has been discussed many times here with (AFAIKT) no real conclusions made. Some people insist that 20-bits cannot be extracted without special hardware. ...

No matter if it's an HDCD or just a marketing trick/special kind of mastering (>20 bit in studio dithered down to 16bit with some special algorithm) - if you copy the CD with EAC you won't lose anything. If you play back the copy on a HDCD capable player (if it is a HDCD) it will be read as HDCD.

To extract > 16 bit audio from a HDCD the only way I know of that could work is using WMP9 on WinXP, a soundcard capable of 24bit digital output and another one (or the same) with 24bit digital input.

I've tried this some time ago with NTONYX Virtual Audio Cable 3.04, but WinXP didn't believe it was a 24bit soundcard. Maybe it will work with a newer version of VAC.

"20- to 16-bit Encoding: Apogee UV-1000 Super CD Encoder"

This doesn't sound like HDCD to me - probably just some extra-expensive mastering equipment has been used, but the CD is still 16bit.


Edit : Removed the quote of the deleted post (HDCS<>SACD)

This post has been edited by Pio2001: May 15 2003, 20:11

[tigre]

http://www.apogeedigital.com/pdf/UV-1000.pdf

...
Unlike bit mapping and noise shaping systems,
it doesn’t attempt to lower the noise level at the
‘most audible frequencies’ at the expense of transparent
operation. Instead, the noise floor is left at the theoretical
16-bit level – but you can hear clear, undistorted detail as
low as 30 dB into the noise.
...
Features & Benefits
• UV22: a unique, industry-acclaimed process
• Allows information beyond the 20-bit level to be
encoded into a 16-bit medium such as CD
• Improves any CD playback system – no decoder
• Completely ‘Red Book’ CD compatible
• Clean audio up to 30 dB into the already low (–96dB)
digital noise floor
• Noise floor is constant irrespective of input signal
• No coloration of the 20-bit source: holds detail,
sound-stage and tonal balance
...

To me this sounds like you have an ordinary audio CD that has been mastered at 24bits (or more) and dithered down to 16 bit using flat dither instead of noise shaped in a final step. I don't see anything special about this. Maybe the ppl who decided to write the remark into the booklet did and gave unwillingly an impression of their knowledge that way.

This post has been edited by tigre: May 8 2003, 14:17

[KikeG]

Apogee UV dithering is a special dithering algorithm that uses as dither signal two bands (U,V?) of noise at high frequencies. Their dither isn't "transparent" (flat?) either, because it increases noise at high frequencies. I don't see any advantage of this over a good noiseshaping dither that also improves dynamic range at lower frequencies.

[tigre]

Apogee UV dithering is a special dithering algorithm that uses as dither signal two bands (U,V?) of noise at high frequencies. Their dither isn't  "transparent" (flat?) either, because it increases noise at high frequencies. I don't see any advantage of this over a good noiseshaping dither that also improves dynamic range at lower frequencies.

If this is true, this

... it doesn’t attempt to lower the noise level at the
‘most audible frequencies’...

is wrong - or not?

If it increases noise at high frequencies (as you say) but doesn't "attempt to lower the noise level at the ‘most audible frequencies’", all in all there's more noise than with plain noiseshaped dither ...

[GeSomeone]

Apogee UV dithering is a special dithering algorithm... I don't see any advantage of this over a good noiseshaping dither that also improves dynamic range at lower frequencies.

It seems more like a special kind of noiseshaping to me:
http://www.apogeedigital.com/news/news_maintain.html

Another technique, ... is "sub-Nyquist dither". This approach, which has been likened to HF bias in an analog tape recorder, adds dither concentrated as near to the Nyquist limit as possible, rendering it inaudible at standard sample rates such as 44.1 kHz and above.

Apogee UV dithering ... uses as dither signal two bands (U,V?) of noise

... UV22 concentrated the dither signal up around an inaudible 22 kHz in a 44.1 kHz sampling environment, hence part of the name (the other part is a reference to "ultra violet", the idea being that it was higher in frequency than Apogee’s signature purple).
To create a carefully-tailored super-audible dither system, Apogee used an algorithm to generate the dither information, rather than relying on noise shaping.

So UV is just part of a marketing name
Anyway it's all about a tool to master 16bit Audio material from 20/24 bit sources.
(Thanks for the link Tigre)
--
Ge Someone

[KikeG]

Still, UV uses two bands of noise at high frequency as dithering. Check for yourself, see the thread I just started: http://www.hydrogenaudio.org/forums/index....t=ST&f=1&t=9192

Analyzing the UV samples there are two "mountains" of noise centered at 19 KHz and 21 KHz.

[Emanuel]

To extract > 16 bit audio from a HDCD the only way I know of that could work is using WMP9 on WinXP, a soundcard capable of 24bit digital output and another one (or the same) with 24bit digital input.

So theoretically it would be possible to playback a HACD using WMP9, and capture the sound with Total Recorder and play it back from wav-files using a 24-bit capable soundcard? A wav-split could then be made from a cue-sheet genereted by EAC. Not really perfect, since you have to manually start the recording i Total Recorder, but perhaps possible?

[KikeG]

To extract > 16 bit audio from a HDCD the only way I know of that could work is using WMP9 on WinXP, a soundcard capable of 24bit digital output and another one (or the same) with 24bit digital input.

So theoretically it would be possible to playback a HACD using WMP9, and capture the sound with Total Recorder and play it back from wav-files using a 24-bit capable soundcard?

I have already done that, but recording the digital output when playing an HDCD disk with WMP9, instead. I just did a little test some time ago.

There was 24-bit data available at the output, but, first, I haven't checked yet if the output of WMP9 is bit-perfect, and second, who knows if the HDCD decoding of WMP9 is a "proper" decode, equivalent to the one of a hardware decoder.

[tigre]

Are you sure total recorder is capable of 24 bit recording?

[KikeG]

Are you sure total recorder is capable of 24 bit recording?

I didn't use TotalRecorder, I just connected the digital out of a soundcard to the digital in of another, and recorded the data at this digital in with CEP.

[Emanuel]

Are you sure total recorder is capable of 24 bit recording?

Haven't tested yet (will do as soon as I get home), but I'm pretty sure it depends on what your soundcard is capable of. If you don't want to use the software (wich of course is an option), you can use the sound board and then there should be no software limitations. I'll get back and confirm this later.

EDIT: quick results posted below.

This post has been edited by Emanuel: May 8 2003, 23:09

[Pio2001]

Other infos about UV22 : http://recpit.prosoundweb.com/viewtopic.php?t=1284

[Emanuel]

After a quick look with a 24/96 audio adapter (M-Audio Sonica Theathe), I found no option in Total Recorder for recording in 24 bit.

Now, I don't have a 24 bit source to record from at this time, but there may still be an solution. Under the menu "Recording Source and Parameters), you may use "Software" as your recording source. If anyone is interested in trying, make sure your sound card drivers are set to 24 bit, 96 khz, hand make sure Total Recorder does not convert to the "recording parameters" in the same dialog box mentioned above. If you try, I'd be happy to read the results.

[Yash]

First of all, i'd really like to apologize for bringing up this old topic, and i'm really a newbie here. Thousand apologies, cos i stumbled upon this thread while browsing thru the FAQ.

I have a few HDCDs, but i don't have a hardware HDCD player, i actualyl found out that they were HDCD when i played them in WMP9.

Later, i decided to back them up, and i just used Nero to CDCopy (back when i knew nothing). As far as i know, Nero would rip (DAE?) it to temp waves and burn them back to the blank CDR.

I played the CDR with WMP9, and it still had the HDCD logo. Does this mean that the copy is still a HDCD?

Could someone help me out with understanding what happened? Honestly, i can't hear a difference with my sound system (TBSC+Logitech Xtrusio), but i'd really like to learn for academic reasons. Thanks.

[KikeG]

First of all, i'd really like to apologize for bringing up this old topic, and i'm really a newbie here. Thousand apologies, cos i stumbled upon this thread while browsing thru the FAQ.

No need to apologize, there's no problem on bringing old threads, and this isn't that old either.

Later, i decided to back them up, and i just used Nero to CDCopy (back when i knew nothing). As far as i know, Nero would rip (DAE?) it to temp waves and burn them back to the blank CDR.

I played the CDR with WMP9, and it still had the HDCD logo. Does this mean that the copy is still a HDCD?

Yes.

Could someone help me out with understanding what happened? Honestly, i can't hear a difference with my sound system (TBSC+Logitech Xtrusio), but i'd really like to learn for academic reasons. Thanks.

It's because HDCD discs are in fact plain 16-bit 44.1 KHz discs. But they have some special information embedded into the 16-bit data that enables to extract 20-bit equivalent information from this 16-bit data, given that there is a HDCD decoder that can detect and make use of it. At the same time, this 16-bit data can be perfectly played as plain 16-bit data when there is no HDCD decoder available. It is quite a tricky system, and if you search at the forums you will find some more information over how it works. However, the exact procedure is not known, since it is not available to public.

So, when you losslessly copy the 16-bit data at the CD, you retain all the special embedded data, and the disc remains to be an HDCD disc.

This post has been edited by KikeG: May 23 2003, 08:06

[Yash]

Thanks a lot for the info!

[Pio2001]

However, the exact procedure is not known, since it is not available to public.

These papers are very detailed. Only the values of the parameters used for DSP, and the binary LSB codes specifications are missing :

http://www.hdcd.com/partners/proaudio/articles.html

[KikeG]

These papers are very detailed. Only the values of the parameters used for DSP, and the binary LSB codes specifications are missing :

I haven't read them, so I can't comment, but what I meant to say is that some parts of the process are not available to public. If it's only some parameters that are not available, I guess that it doesn't have much importance in order to know how it works.

[Carlos G]

Would it be possible to "reverse engineer" the gain scaling part of HDCD, using WMP9 as the"black box" to test, so that it could be implemented in software decoders?

The quotes are from the "Gain Scale" paper in http://www.hdcd.com/partners/proaudio/articles.html .

The "mark" to activate gain scaling seems to be only one bit of the sample:

The Digital Gain Scaling solution requires 1 bit of DAC resolution to properly decode Peak
Extended recordings. This means that on non-HDCD recordings, the PMD-100 IC loses 1 bit of
resolution.

By trial&error, maybe is possible to find the right codes for the LSB. You feed WMP9 with test signals, until it activates the HDCD decoding. Or you feed WMP9 with a high amplitude signal, changing the LSB and try to find which changes of the LSB make an increase of the dynamic range. The codes are present in HDCD PCM streams, so by analysing them maybe it is possible to get a little help...

Peak Extend is a restorable, instantaneous soft peak limiter which has a direct one - to -one mapping so that it can be easily and quickly restored in the HDCD decoder. Because this operation is performed digitally, it has a precise and stable curve

So if there's a 1-to-1 relationship between input and output, it is possible to obtain the exact transfer funcion...

Then software decoders could convert the 16 bit HDCD signal to a 24 bit, range extended signal...

Carlos G.

[mundschuss]

so, its quiet easy to extract the 20bit audio from HDCD:

you just need "dBpoweramp" and download the DSP extensions. activate "HDCD"-DSP PlugIn and dBpoweramp will save a 24bit file. if theres no HDCD in the drive, he'll save it as 16bit.

i currently try it out with tool's lateralus...


seeya