If you reckon if they'd standardized the hugely complex elliptic A-A filters that were generally in use in those days (early 80s) we'd all be using DSPs applying their phase inverse to 'correct' the rather more benign phase characteristics of the digital filters used today? Or that those analog filters would have gotten set in stone preventing advances to digital ones? Either way doesn't sound like a recipe for great sound to me.
If it's standardized, the pre-correction could be done at the recording end, even with early-80s technology. I put a demonstration together at the time for the management of the company I worked for, using our proprietary computers (which is what we sold); unfortunately, although it was easy to see the advantages on a scope, none of the big bosses could hear any difference. As much as I tried to convince them that whether or not they heard any improvement was a different matter than whether audiophiles would believe that they heard improvements, the project was killed. Ah, well.
Yes and if standardized it would still be built into the decoding chips even today, meaning that nobody could use the technically superior digital filtering which came along later without listening through inverse-cauer phase correction.
<edit> Brain fart - sorry didn't read your posting carefully enough! So the recording equipment could be upgraded without needing any corresponding upgrade to the replay equipment then? So why would it need to be standardized? Surely the A-A filter and the pre-correction of phase could just be a single unit?
<edit> Brain fart - sorry didn't read your posting carefully enough! So the recording equipment could be upgraded without needing any corresponding upgrade to the replay equipment then? So why would it need to be standardized? Surely the A-A filter and the pre-correction of phase could just be a single unit?
Last edited:
After re-reading a couple of rsdio's post I realize that I should address one point. I am simply stating that no resampling/anti-imaging algorithm that I have seen used currently even approaches the numerical limit of 64bit floats.
It is true that a real music signal sampled at 192kHz will have energy in even a 20-22.1kHz transition band. I would like to see the best a brute force approach could do on real signals. I won't volunteer to hear the difference since I can't hear a 16bit 17KHz brickwall.
It is true that a real music signal sampled at 192kHz will have energy in even a 20-22.1kHz transition band. I would like to see the best a brute force approach could do on real signals. I won't volunteer to hear the difference since I can't hear a 16bit 17KHz brickwall.
Fortunately for us oldies, at 17kHz it don't need to be a brickwall at all. There's a full 5kHz for the transition band
So you ran some digital tape machine at 1/100th normal speed so it was correctly sync'd up to the uber-slow computer? Who did the engineering to put the tape machine into action replay mode? Sounds like a fun job
No, we didn't use tape- we had large hard disks. You could get almost 5 minutes of music on them! The signals were processed and dumped onto another set of hard disks.
They were flying saucer sized. Had handles.
Just to give scale, the instrument I was working on at the time was about the size of a Mini Cooper, with a full desk console and an equipment rack. I now have in my lab a functional equivalent, albeit much easier to use, faster, more versatile. It's the size of a small printer and cost 10% of the earlier unit's price in actual dollars; accounting for inflation, I suppose that would actually be closer to 3%.
Damn, I starting to feel as old as John.
Just to give scale, the instrument I was working on at the time was about the size of a Mini Cooper, with a full desk console and an equipment rack. I now have in my lab a functional equivalent, albeit much easier to use, faster, more versatile. It's the size of a small printer and cost 10% of the earlier unit's price in actual dollars; accounting for inflation, I suppose that would actually be closer to 3%.
Damn, I starting to feel as old as John.
As an example of the brute force approach, take an entire piece of music as one mass FFT zero the bins past Fo and do an inverse FFT. That way you use all of the information no more no less. Or run a properly overlapped 2^24 point FFT filter over the data. In both cases a HUGE latency for any current computers (home style).
Actually, the old analog synthesizers such as the Moogs, which employed subtractive synthesis, used square, pulse, triangular and sawtooth waveforms as the sound source.
se
A clever and open way to circumvent standardization and still provide data would have been to store the impulse response and additional info on the CD in a hidden track or subcode etc, that is, the impulse of the recording chain (last mastering stage, or earlier if neccessary, for example when minimum-phase multiband compressors were used -- which are allpasses like most crossovers).
That would allow more advanced playing gear to make use of the information. For computer-based components with digital data/audio interconnects it likewise would be useful if eg USB/FireWire-DACs would publish their impulse reponse(s) to the host as well.
The only thing standardized in the Red Book in that regard is the emphasis flag, to my knowledge.
- Klaus
Well, most of what I did that was really educational happened about 35 years ago, +/- 10 years. That is when I was younger, more flexible, and better funded. Now, I am a tired old man, almost 5 years into retirement, yet I'm now designing 3 major products, and have another 5 or 6 out there in the marketplace. Yet, I must agree, that the best place to learn 'new' information pertaining to audio quality is the links provided by others on this thread, and reading them through. I am certainly not going to write another 'breakthrough' paper or do all the research necessary to publish in the IEEE or JAES, and still have it ignored by the majority of the armchair critics out there. However, I do want to encourage others, reading this thread, to do the latest research and link it here. I certainly appreciate and will give it due consideration.
Now, what amazes me is the rather lowish level of understanding of digital problems shown by many people here, that we knew about 35 years ago or more! Swept it all under the rug, have you? Not on my watch! '-)
Now, what amazes me is the rather lowish level of understanding of digital problems shown by many people here, that we knew about 35 years ago or more! Swept it all under the rug, have you? Not on my watch! '-)
Well, 40 years ago, I was having lunch with Richard Heyser.
39 years ago, I was having dinner with Bob Stuart
38 years ago, I was having dinner with Mark Levinson
37 years ago, I was having dinner with engineers at Ortofon.
37 years ago, I was having lunch at Philips Research Labs as they showed off their
39 years ago, I was having dinner with Bob Stuart
38 years ago, I was having dinner with Mark Levinson
37 years ago, I was having dinner with engineers at Ortofon.
37 years ago, I was having lunch at Philips Research Labs as they showed off their
Scott Wurcer and I had pizza together about 25 years ago. Kind of a new 'kid' wet behind the ears, at the time. He actually got along with me back then. Of course, in that period, there was Dr. Vandenhul, Dr Hawksford, and many others, who visited my little lab and we talked together at some length. Some of us really like audio design and we love to talk about it.
- Status
- Not open for further replies.