DVV . Even the MP3 can sound good . A look at DAC's , buffers, clocks , and with and without final filtering perhaps . I think it is time I built a universal digital interface for myself .
The Meridian 14 bit CD players still to me had more magic than the others . They were highly praised in their own time . I wonder how many kept them when the CD discs improved ? They would have been very surprised . Maybe a little less detailed than the very best , like taking a slightly less good seat at the concert hall .
Even the best digital I suspect does not investigate the sub 1 watt ?
The Meridian 14 bit CD players still to me had more magic than the others . They were highly praised in their own time . I wonder how many kept them when the CD discs improved ? They would have been very surprised . Maybe a little less detailed than the very best , like taking a slightly less good seat at the concert hall .
Even the best digital I suspect does not investigate the sub 1 watt ?
That I have yet to hear, MP3 without the bloated bass and weak treble.
But I'm a patient man.
I've done some interesting experiments of taking nominally fairly squalid MP3 files and resampling in Audacity up to the level of very high resolution files. MP3 -> 24/384 ... pretty extreme, huh, but it works! In terms of "fixing" the SQ, that is; at least, on playback via the very ordinary PC, on mainboard chip and conventional computer speakers.
Downsides are, very lengthy processing time in Audacity, and monstrous output file sizes ...
Why does this work? My theory is that a lot of the degradation is caused by the normal, real time processing of the MP3 file; if you offline the process then you get much, much better results.
Frank
Downsides are, very lengthy processing time in Audacity, and monstrous output file sizes ...
Why does this work? My theory is that a lot of the degradation is caused by the normal, real time processing of the MP3 file; if you offline the process then you get much, much better results.
Frank
Fair enough, I won't argue that technically there are inadequacies, perhaps other algorithms can do an even better job. My point, though, is that, in my experience, the resampling made the world of difference in the quality of replay of an ordinary MP3 file, which to me is purely a result of the CPU not having to decode the MP3 file in real time.
In particular, the quality of treble dramatically improved. A cymbal strike in the original MP3 was just a bit of dull, white noise; the resampled, hi res output was a hell of a lot closer to sounding like proper instrument shimmer ...
Frank
How mosquito with elephant eyes can be better than an ordinary mosquito? Better engineered = more optimal for the functions.
I tried, from memory, 3 different players. Including foobar2000. Made slight differences to replay of MP3, but nowhere to the degree that resampling to different rates did. Audacity was never used for replay, merely for the processing (blown sound chip on work m/c).
There was a very precise, stepped improvement in quality with each upping the ante: MP3 -> 16/44.1 -> 24/96 -> 24/192 -> 24/384. Each time, resampling from the original MP3, and creating ever larger files. The last round took about half an hour to process, for a normal pop track!
Again, I think the gains were that the CPU and sound chip were having ever easier times during replay, and less electrical thrashing means better sound quality ...
Frank
Better sounding. I thought that was fairly obvious given the context myself
Show reasoning based on the evidence of my postings here or elsewhere. The conclusion is incorrect so you must have jumped to a wrong delusion or two somewhere. I'd like to help you straighten that out.
Over-engineered results in a sub-optimal use of engineering resource - including customer money. You were previously unaware of that?
Yes, perhaps its the sound chip - the digital filter used will be different if it actually does operate at a different frequency (not by any means certain with Windows - which has a tendency to resample - in the way).
Wave, in the theory of cybernetcs, there are no multiple optimal points, there is only ONE optimal point, all others are not optimal. Therefore, there can be no "more" or "less" optimal, all but one are non-optimal. For further info, please consult Wiener's "The Human Use Of Human Beings" and the works of Oscar Lange, who will explain this in mathematical terms.
Think of it this way - think of yourself as a unique event in the cosmos. There is only one Wave, all else is sub-Wave.
Half an hour? Wow! That becomes some serious time when the number of tracks goes up to 800, or as much as my current USB 4 GB memory stick holds of WMA tracks.
The stick is used in my car only, a recently purchased Chevrolet Cruze. Although it's equipped with a standard CD player/radio, and can take standard CDs no problemo, from USB it will take only compressed signals, three of which I don't even recognize, but WMA, the Windows media player format, I do recognize and can do. No idea why it will not accept standard WAVE format from the stick, but will from the CD.
And, of course, would require me to buy quite a few 32 or 64 GB memory sticks for the same 800 titles.
Did you by chance keep records of the time required to bump the standard 16-bit WAVE format to higher levels? Just curious ...
There are degrees of sub-optimality. There are also local optimums which are sub-optimal when the bigger picture is considered. In numerical optimization algorithms, its assumed the starting guess is not optimal and the algorithm will proceed in the direction of greater optimisation. So if you stop the algorithm along the way then you'll get a 'less optimal' solution. I'm sure Wave is not suggesting that there are multiple optimal points so it does rather look as though you're tilting at windmills.
Norbert Wiener's a cool guy though
<edit> perhaps I should have used 'optima' instead of 'optimums' ?
Norbert Wiener's a cool guy though
<edit> perhaps I should have used 'optima' instead of 'optimums' ?
There are degrees of sub-optimality. There are also local optimums which are sub-optimal when the bigger picture is considered. In numerical optimization algorithms, its assumed the starting guess is not optimal and the algorithm will proceed in the direction of greater optimisation. So if you stop the algorithm along the way then you'll get a 'less optimal' solution. I'm sure Wave is not suggesting that there are multiple optimal points so it does rather look as though you're tilting at windmills.
Norbert Wiener's a cool guy though
<edit> perhaps I should have used 'optima' instead of 'optimums' ?
I never questioned degrees of suboptimality.
The rest in no way changes the fact that there is only ONE optimal and theoretically an infinite number of suboptimal points (never really true, the number of possible solutions which will not degrade the problem further is usually a very finite number).
Stopping the algorithim before completion will simply produce another suboptimal point, no more, no less.
All of this does not change the fact that there can be no "more optimal" or "less optimal" point, optimum is a single value quantity which cannot be stretched to "more" or "less". One can have a number of solutions which will be nearer or further from the optimal point, but that's it.
As for "local optimums", in the end, they will dance to the tune of the overall system optimum, won't they? What do we care more about, parts or the whole?
That last sentence is what I mean by 'less optimum' - further away from optimal. I took it that Wave meant the same.
But in engineering we do have to erect notional boundaries to circumscribe what's considered 'the system' and what's not, even though we know we've put them there ourselves. Considering solutions 'optimum for the whole universe' is a bit too mind-boggling. How do we know when we're considering the whole enchilada?
I often found good mono gives the illusion of being stereo . The depth is a great substitute and is not unlike real life classical music . I must try some MP3 mono to see if it has similar effects . I feel it must albeit about depth .
One guy came up to me when I was talking about Garrard 401 at a show . He said he uses a 401 to check digital systems as he then has something like a live performance that can be called upon to work whenever he wants . Often he said the strong voice of the Garrard went missing . More obvious than trying to remember how something sounded that was live .
I suspect an analogue master tape might not be as good . It is what is slightly wrong with a Garrard that should not change . It does . It is as if the Garrard had a stroke .
As I said before an SACD of Lad Zeppelin sounded like a tribute band against the original , the acetate test cut sounded great.
MP3 I like as it confounds my expectations . I suspect I like it more than what should be better . I remember years ago someone said digital and class B had subtractive distrotion ( loss or granular ) . Valves and analogue additive . The latter being more acceptable in small amounts .
In engineering especially, the overall optimum may, and in reality very often does, mean that for the sake of achieving the optimum point for the whole, subsystems may have to be purposely moved away from what may be their own subsystem optimum.
Since engineering does not live in a vacuum, price is always a consideration. At any and all points, price will eventually come up as an important selling point, and will cause the product, if it is to achieve its optimum (selling point) from the perspective of its makers, must also satisfy various price, as well as technical criteria.
This may cause the manufacturer to quite intentionally compromise the technical optimum for the sake of achieving and overll sales optimum, because, let be frank about this, they are in it to sell.
A good real life example - my own new car. It's relatively heavy (1,365 kg, or about 3,000 lbs), so it's no brainer that it would benefit from an engine which would not trade horsepower for torque. Consequently, from its roster of engines, the ideal (for normal driving) engine would be the 1.4 litre turbo, as it provides 140 hp and 200 Nm of torque at 1,800 rpm, rather than the installed engine of 1.8 litres, 141 hp but only 176 Nm of torque at 3,500 rpm. Technically speaking, the turbo is a better choice as overall, it also enables a better fuel consumption figure.
That model is available on the US and Australian markets, but not in Europe. If it were available in Europe, very few people would choose to buy an Opel Astra, which is essentially the same car but WITH turbo engines available, and at a price of some €5,000 more, or at a 33% higher cost. The more expensive Astra is not selling very well in Eastern European markets, whereas the Cruze is doing well.
Thus, in this case GM made a conscious choice to use a suboptimal subsystem to achieve the overall optmial goal of good sales.
Since I own the suboptimal version, having no choice, I can safely say that for normal, everyday use as a family car, the lack of optimum is not really a problem, and given where I live, it is in fact perhaps a better solution overall since it's a simpler product, much easier to maintain.
But, despite my personal satisfaction, obviously subject to my own taste and requirements, the compromise in favor of the price and practicality is a good one, even if technically speaking, this is the suboptimal version.
I agree, a good mono recording can contain a surprising amount of spatial information in all three dimensions: Up and down, back to front and (albeit limited) left and right.
A good stereo recording, regardless of medium, should greatly expand the left to right spread without diminishing the other directions.
To my ears most MP3s largely lose all of that except the stereo mockery of hard left-dead centre-hard right I mentioned earlier leaving no real depth either.
- Status
- Not open for further replies.