Software decoding

[surroundophile]

I don't think I've ever seen it discussed, but how does software decoding compare to the Tate's decoding in regards to:
1) Image (position, channel leakage, pumping, ambience)
2) Fidelity (sound quality, highs, lows, distortions)
3) Cost (what software is needed)

Thanks

 

[sq10]

I don't think I've ever seen it discussed, but how does software decoding compare to the Tate's decoding in regards to:
1) Image (position, channel leakage, pumping, ambience)
2) Fidelity (sound quality, highs, lows, distortions)
3) Cost (what software is needed)

Thanks

Yes, excellent question....thanks!

 

[neil wilkes]

How long is a piece of string?
Seriously - these type of "X Vs Y" threads are about as meaningful & useful as a chocolate teapot.
It is all so subjective, and with any computer based audio systems the more you spend the better it will get - just like in the real world.
The actual scripts do a superb job, but yet again will be entirely dependant on the SQ or QS stream you're using, which is dependant on the condition of the vinyl, the quality of the cartridge & stylus used, the quality of the preamps used, the quality of the Phono amp used etc.
Done properly the script has the potential to be far better than the Tate. No worn components, no noisy degraded caps, but you get my point now I think.
Software needed is Adobe's Audition.

 

[Bob Romano]

Neil is correct. This question is pretty subjective.

I use the scripts exclusively and get excellent results usually. I have never used a QSD-1 or Tate to do a conversion but I have heard a couple hundred or so using them and then I have done many conversions of the same material using the scripts. To my ears, they sound every bit as good. The scripts are time-consuming but simple to use and all you really need is a turntable, the vinyl and Adobe Audition. There are scripts for every version from AA1.5 to the present 3.0. They all do pretty much the same job.

The difference for you may be your cartridge, your amp setup etc. Obviouisly, the cleaner your sound path and the better conditon your vinyl is in, the better your outcome.

If you are goin to make a disc out of them for playback on your home system or in your car you also need a few other programs.

Now where is that chocolate teapot??? I am hungry!!!

 

[Bob Squires Jr]

Leave it to the English to invent the chocolate teapot!
(Meanwhile, back in the states... we're still trying to figure out what to add to powdered water )

I've used the scripts, a Tate unit and a Sony SQ decoder for my converts,
and so far the scripts seem to be the best, followed by the FT II.
As Neil stated earlier, there are no problems with the scripts like dried caps
and crackly transistors etc. so the decode should, in theory, be pretty much dead on.


-B

 

[Lucanu]

We're speaking of course about conversions, that is to obtain the best decoding possible to preserve both record and obtain the original 4 channels as it was intended to be heard and software (scripts for Audition) are the best for this, takes a long time to decode properly (depends on how powerful is you pc) but it restores almost prefectly the original mix, and as the software/hardware improves we can hope to obtain almost perfect recovering of the original 4 channels!

 

[Key]

I think the weakest link for both SQ and QS is the 90 degree phase shift. There are different ways to do it in software and they both have there up sides and downsides.
Do any of these scripts use a control signal, attack time, and release time? The tate uses a logic system to sort of switch between 2 matrix decoders really fast as far as I can tell.

 

[ingresman]

Firstly for our American friends, we also have chocolate fire guards in the UK, but as no one can afford fuel anymore they are soon to be relegated to the bin. Perhaps in 39 years someone will start www.chocoloatefireguardchocolate.com !
Also assuming 1 gallon = 4.55 litres, we pay rought £1.10 per litre so thats about £5 per gallon or roughy $10 - how much does anyone else pay around the world ????
but to be serious....
i'd be interested to see an alternative to doing 90degree shift in software other than a hilbert transform, if someone could post I'd be gratefull.
The willcox patrent http://www.freepatentsonline.com/3944735.pdf has a host of goodies in it.
I've read it about 50 times and each time I get a bit more into, the maths are way beyond me. What I get from it is essentuialy each channel is made up of its original content + other channel information at various phase shifts and applitude that cause crosstalk.
What you have to do for each channel is see what shouldn't be there and remove it and at the same time make sure that the total power output from all 4 channels remains the same as the original level (I suspect if this doesn't happen correctly you get the classic "pumping" of logic decoders)
To implement in software would be possible as it's all basic maths on a sample at the end of the day, it would take time to execute as you have to apply some quite involved maths to each sample . This is the what the detector does in the original patent.
If someone could explain how to do the identification I can probabbly provide the coding skills.
Interesting the patent talks about a different set of values fo SQ and QS so it would be possible to set up a common "logic" module to handle any basic decodes from both systems.
I've previously written about the developent of a basic SQ decoder (with some blends from the mc1312 data sheet) which I can resurect if any one is interested.

 

[Key]

i'd be interested to see an alternative to doing 90degree shift in software other than a hilbert transform, if someone could post I'd be gratefull.

I'd like to point out that I am in no way a DSP expert - I am an infant when it comes to these things - coding dsp and dsp practical theory. But I will try my best to explain a little about what I know and maybe someone can correct me where I mess up if they read this and know better.

Implimenting a hibert transform isn't exactly straight forward. In the first place you can't do an actual hibert transform because it is "non-casual". There are different ways to practically or approximately do it over the full frequency range (20Hz - 20kHz) - and they are all imperfect imo. The 2 basic filtering methods used are FIR and IIR filters.

I think one method is to model multiple FIR band pass filters that approximate the hibert transform and then to combine the multiple bands at the end of the transform - this like other methods causes phase and frequency distortions which then must be offset before the signal reaches the output. The idea is to mitigate the phase and frequency distortions as much as possible.

An example of a program already on the market that utilizes both FIR and IIR for phase shifting is the VST/DX plugin Ultrafunkhase.

 

[Lucanu]

In my latest SQ script/session I did use an external phase shifter VST plugin called PhaseBug. It would be nice to try to change that plugin with another into the session file and see if there is any difference in results

Good point Key!

 

[neil wilkes]

There is a wonderful Phase Correction plugin (I do not know if it is the kind required here, but if someone can email me with the exact requirements I will ask the coders) called PHA979 by Voxengo.
Sadly it is not a free plugin though - would this matter?

 

[Lucanu]

Sadly it is not a free plugin though - would this matter?

....eeer.....

 

[neil wilkes]

I'll take that as a "Yes, it matters" then.
Back to the old drawing board....

(BTW - I have nicked your idea of adding a Google Earth location to my sig.
Great Idea!)

 

[Lucanu]

Thanks, it's nice to "see" were we live or work, it gives me the feeling we're closer each other

 

[ingresman]

Neil,
that's a kind offer. I think all we need to start is the 90degree shift which I belive is where the hilbert comes in. It gets implemented as a series of FIR or IIR fillters I belive (can't remember which)
How much would the plug in cost, it might be a lot cheaper than audition.
Would it be possible to ask them what there view is on how to get a shift, I would think thier DSP knowledge is better than mine (which ain't saying much).
What becomes more interesting is how to implement something like a 60degree shift which would be needed to do a matrix H decoder in software.