Quantcast

Working CD-4 (software) Demodulator!

Help Support QuadraphonicQuad:

winopener

2K Club - QQ Super Nova
Since 2002/2003
Joined
Mar 2, 2002
Messages
3,836
I do think it's a great idea, and that maybe the expertise of someone here can help them sorting out the cd4 quirks.

Some thing i remember from years of reading here...

1) cart and preamps: i don't know the specific models they had used: especially for the cart, anyone is using it on a real cd4 setup? It may be excellent, but in order to build a test bench to run a new software, all the other elements had to be 100% ok for the specific task. If the cart they are using can track and read without distorsion the Cat Stevens Greatest Hits USA CD4 it's ok - it's probably the hottest cd4 ever done.

2) since they do 4x oversampling of a 96k sample rate (thus working at 384), maybe better to sample at 192 and do a 2x oversample? The a/d converters would not work in the border limit zone or risking to cut out FM information.

3) noise reduction: cd4 rear channel implements ANRS for noise reduction. That's quite a beast to do in software, as with Dolby.

4) time alignment: i know it has been discussed here years ago that there's a slight delay between front/rear to compensate for the big differencies in the signal path. It is probably the most critical point in order to make it work the sum and difference matrix. A real CD4 lp with a silent channel for a long time can be very useful in fine tuning this aspect, the first that come on memory is the japanese CD4 version of Deodato Prelude which has on track 1 the rear left channel totally silent for a long time. Since it is a quite rare title, any alternative is welcomed.

BTW, CD4 records can be decoded excellent also in 2018...
 
Last edited:

ArmyOfQuad

2K Club - QQ Super Nova
Since 2002/2003
Joined
Apr 22, 2002
Messages
2,047
Location
Attleboro, MA
I've been exchanging emails with them. This is getting interesting.

I missed the mac only part of this though.....I never thought I'd say this....I might need to buy a mac now. Ugh.
 

winopener

2K Club - QQ Super Nova
Since 2002/2003
Joined
Mar 2, 2002
Messages
3,836
I feel your pain... from the mac userbase, it is still possible to repurpose a intel-windows machine into a intel-mac? Since this is going quite offtopic, please just some link to follow, not detailed explanation here.
 

kfbkfb

400 Club - QQ All-Star
QQ Supporter
Since 2002/2003
Joined
Feb 20, 2003
Messages
421
Location
Midwest USA
Latest info at the link in the first post:

Stereo Lab will support software decoding of CD-4 quadraphonic LPs.
The timescale for the integration in the software is not yet decided.
However, the concept has been proved and we are ready to code this
when the next development slot becomes available.


(considering buying a Mac just for this)

Kirk Bayne
 

The Quadfather

1K Club - QQ Shooting Star
Since 2002/2003
Joined
Mar 7, 2002
Messages
1,569
Location
Dixie
3) noise reduction: cd4 rear channel implements ANRS for noise reduction.

The ANRS is not implemented in the rear channels, it is implemented in the difference channels before they are mixed in with the main channels to get the four discrete channels. So the ANRS is in all four channels.
 

winopener

2K Club - QQ Super Nova
Since 2002/2003
Joined
Mar 2, 2002
Messages
3,836
The AWB Pick Up The Pieces shows the limits of the intermediate Ambisonic approach, front-back bleeding as it was a half-logic matrix system. If they can eliminate the intermediate Ambisonic step and decode directly as CD4 was intened, i'm in.
 

ArmyOfQuad

2K Club - QQ Super Nova
Since 2002/2003
Joined
Apr 22, 2002
Messages
2,047
Location
Attleboro, MA
I think it may be interesting to see how this evolves, that they've gotten these results in their early tests shows promise. I do rather like their intent for it to "fail gracefully", which could come in handy for problematic tracks.

For now I'm waiting to see how this evolves. But perhaps a decoder with options could be a nice final product.
 

kfbkfb

400 Club - QQ All-Star
QQ Supporter
Since 2002/2003
Joined
Feb 20, 2003
Messages
421
Location
Midwest USA
Maybe 2 Quadraphonic output files -
CD-4 decoded the old way with the pure demodulated F-B signal
(even if it has poor sound quality) with F-B muting only on carrier loss
and
CD-4 decoded the new way with Ambisonics

Added:
I brought my headphones to the Library PC, the "Listening Room"
stereo audio segments sounded excellent, it's amazing to hear an
all software CD-4 decode.

Kirk Bayne
 
Last edited:

Brice

Active Member
Joined
Jun 13, 2018
Messages
54
Location
France
All,

As the Pspatial Audio CD-4 decoder is being discussed here, I thought it was time to join the forum because I am responsible for the decoding algorithm development. Let me know if there are any questions I can (try to) answer.

There has been some discussion regarding the "soft fail" approach that I have taken. I do agree that (inevitably) this will sacrifice some degree of the strong directionality of CD-4. (I also know that it is this aspect which is so admired by quadraphonic fans.) Nevertheless, within the limitations of the Stereo Lab software program which is a "fire and forget" decode, I need to be sure that the decode will be be reasonably reliable. Because of this, I have to "play it safe" and aim to get a reasonable decode in most circumstances.

For the present, I'm not going to try to emulate the best CD-4 decode from the best hardware. I am assuming that the discs, and the associated hardware, may not be "tuned" to CD-4 performance. In this way, at least, Stereo Lab should complement a top-flight hardware system.

Kirk's point that we could offer a pure decode is a distinct possibility. In fact, it involves a lot more processing to take the approach I have. Straightforward emulation of a 1970s hardware decoder is absolutely possible. We can look at this at a later date if folks here are interested for us to do that.

May I make a request for any needle-drops of CD-4 material please. But there's a catch, I do need a 96kHz file UNEQUALISED. Post RIAA eq reduces the carrier signals too much. If anyone has the equipment to provide needle-drops of this type, I would be very grateful. I'm getting very bored of the few CD-4 discs that I have! :)

Best,

Richard Brice
 

winopener

2K Club - QQ Super Nova
Since 2002/2003
Joined
Mar 2, 2002
Messages
3,836
Hello Richard,
first of all, thanks for stopping by here.

96K Unequalized means to connect the cart directly to a linear preamp, not a phono one, in order to raise the signal and then sample it at 96k. Correct?
 

Brice

Active Member
Joined
Jun 13, 2018
Messages
54
Location
France
Hi Winopener,

Thank you for the welcome.

Yes, exactly as you say; the cartridge level needs to be increased (to get a good signal to noise ratio) but the EQ needs not to be applied. Hardware decoders all had the carrier signal take-off prior to the RIAA de-emphasis and this has proved to be important for us as well.

Best,

Richard
 

Brice

Active Member
Joined
Jun 13, 2018
Messages
54
Location
France
Hi,

Yes, just watch out that mic' preamps tend to have a low input impedance (2k). That's too low for a MM cartridge.
 
Top