'Data' and 'theory' don't even attempt to account for the 20-30 db of *transient* *spatially localized* information that the human hearing system can reliably process below an acoustically random noise floor, as a result of the evolutionary process, and that analog generally also can process.
A signal analyzer, given enough time -probably hundreds of cycles worth and with appropriate windowing, can winnow out continuous single frequency signal information below an appropriately dithered noise floor, to which I say - what realistic correlation does that have to do with the human hearing perceptual mechanism, especially in the time domain? Too much meter beating going on here, IMO.
I could probably design a 1 bit digital processor from which I could extract a sine wave 120 db below the 1 bit level with the appropriate setup. So what?
A signal analyzer, given enough time -probably hundreds of cycles worth and with appropriate windowing, can winnow out continuous single frequency signal information below an appropriately dithered noise floor, to which I say - what realistic correlation does that have to do with the human hearing perceptual mechanism, especially in the time domain? Too much meter beating going on here, IMO.
I could probably design a 1 bit digital processor from which I could extract a sine wave 120 db below the 1 bit level with the appropriate setup. So what?
Last edited:
Thor, go listen to the samples that Werner Ogiers posted in his dither article. Use your ears.
There is nothing that analog can process that a digital system can't. 16 bits for playback absolutely blows away anything that SOTA analog tape or LP systems can do. It's not even close. The signal goes cleanly below the noise floor, and the noise floor is an order of magnitude better than the quietest room you'll ever sit in.
Do what I did if you want to woffle about human hearing- compare by ear a direct mike feed to the same with a A/D-D/A inserted in the chain. Analog, in any form, is a dead issue except for niches that are driven by other criteria than accurately reproducing what microphones can capture.
There is nothing that analog can process that a digital system can't. 16 bits for playback absolutely blows away anything that SOTA analog tape or LP systems can do. It's not even close. The signal goes cleanly below the noise floor, and the noise floor is an order of magnitude better than the quietest room you'll ever sit in.
Do what I did if you want to woffle about human hearing- compare by ear a direct mike feed to the same with a A/D-D/A inserted in the chain. Analog, in any form, is a dead issue except for niches that are driven by other criteria than accurately reproducing what microphones can capture.
Sy -
As a designer of digital systems, I'm not going to quibble about best case DSP processing ability, regardless of the fact that many such systems in reality have been sadly deficient here largely based on the 'digital is perfect' mantra - it's conversion techniques to unrobust standards and related artifacts that are the primary problems I am concerned with. Many people who internalized Sony's CD advertising hype were screaming that 16 bit digital was 'perfect' as well for professional work for decades before the recording industry decided to ignore such opinions based on what they were hearing.
Not sure why you would attempt to deny the existence of brick wall artifacts and quantization distortion which runs at about 0.25% at a full 8 bits of resolution and near 50% at the least significant bit. Amelioration schemes for this fall short of the panacea claimed with real life signals, as I've pointed out, and are applied only in a minority of systems IAC.
Arbitrary claims such as
As a designer of digital systems, I'm not going to quibble about best case DSP processing ability, regardless of the fact that many such systems in reality have been sadly deficient here largely based on the 'digital is perfect' mantra - it's conversion techniques to unrobust standards and related artifacts that are the primary problems I am concerned with. Many people who internalized Sony's CD advertising hype were screaming that 16 bit digital was 'perfect' as well for professional work for decades before the recording industry decided to ignore such opinions based on what they were hearing.
Not sure why you would attempt to deny the existence of brick wall artifacts and quantization distortion which runs at about 0.25% at a full 8 bits of resolution and near 50% at the least significant bit. Amelioration schemes for this fall short of the panacea claimed with real life signals, as I've pointed out, and are applied only in a minority of systems IAC.
Arbitrary claims such as
are short of context in reality, unless you believe real time distortion figures exceeding 10% are 'clean'. Plus, as I have pointed out even band limited 24 bit PCM systems suffer from envelope distortion at higher frequencies of up to several percent which is why DSP engineers concerned about signal envelopes feel it necessary to use a sampling rate of dozens of times the frequency of interest. Feel free to deny any research about human hearing you want - that only stengthens my case.
Last edited:
I guess you didn't bother to read the article on dither and actually listen to the samples. If you'd rather believe your eyes rather than your ears, look at the spectral data that Stereophile presents at low signal levels for their 16 bit player reviews.
This was the stuff of audiophile hysteria 30 years ago.
First CD player that popped up in Stereophile reviews.
Parasound Halo CD 1 CD player Measurements | Stereophile.com
See Figure 4. At -90dBFS. Can you show me the horrific quantization distortion? The only harmonic I can see is the second, at better than -125dBFS, lower than the hum.
Second one to pop up:
Audio Research Reference CD9 CD player/DAC Measurements | Stereophile.com
See Figure 9. No harmonics visible, just an idle tone and some noise waaaaay below any reasonable threshold.
Yes, there are also poor performers, but that's not because of the format, it's because of the engineering, such as it is.
Parasound Halo CD 1 CD player Measurements | Stereophile.com
See Figure 4. At -90dBFS. Can you show me the horrific quantization distortion? The only harmonic I can see is the second, at better than -125dBFS, lower than the hum.
Second one to pop up:
Audio Research Reference CD9 CD player/DAC Measurements | Stereophile.com
See Figure 9. No harmonics visible, just an idle tone and some noise waaaaay below any reasonable threshold.
Yes, there are also poor performers, but that's not because of the format, it's because of the engineering, such as it is.
For $1300 to $4500, this performance would hopefully be what is expected from a competently designed CD player. But how about the average $150 Best Buy player that the '99%' own? None of this does anything for a non- properly dithered source, anyway, so what's the point?
The fact is that I can readily hear the differences between standard CD, the relative improvement on noise-shaped CD and 24 bit audio (all of which I have sources of) consistent with their standards on my second best (all SS) system, which none of you are demonstrating, with your systems. I don't see why the implication is that I'm not hearing any difference.
Plus, I've demonstrated the possibility of an unprecedented quality analog recorder /player based on existing technology which indicates that I am not 'denying' anything.
The fact is that I can readily hear the differences between standard CD, the relative improvement on noise-shaped CD and 24 bit audio (all of which I have sources of) consistent with their standards on my second best (all SS) system, which none of you are demonstrating, with your systems. I don't see why the implication is that I'm not hearing any difference.
Plus, I've demonstrated the possibility of an unprecedented quality analog recorder /player based on existing technology which indicates that I am not 'denying' anything.
Last edited:
Have you? All you require for your 'possibility' is unachievable analogue linearity in FM generation and discrimination, combined with unachievable dropout performance for any realistic medium. Digital techniques could do it, but wouldn't they spoil the fun of a new analogue recording/distribution technique?thoriated said:
Don't see why a digital BIT function couldn't trim out an all analog circuit when no signal is present to any desired precision, then disable itself at other times. Result: pure analog magick
IAC, even an all-analog approach w/o BIT should easily beat out the performance of a high quality FM tuner by probably an order of magnitude due to the relative insignificance or lack of signal strength differences, multipath, noise interference, HF carriers, etc.
Hey - I'll be happy to design this if somebody will pay me living wages and NRE.
IAC, even an all-analog approach w/o BIT should easily beat out the performance of a high quality FM tuner by probably an order of magnitude due to the relative insignificance or lack of signal strength differences, multipath, noise interference, HF carriers, etc.
Hey - I'll be happy to design this if somebody will pay me living wages and NRE.
Last edited:
You will still get signal strength differences due to read head flying height.
You may get interference from adjacent tracks or symbols, unless you are happy with recording density well below start of the art.
A high quality FM tuner is not only limited by the things you mention. If you have ever designed one, you might know.
You may get interference from adjacent tracks or symbols, unless you are happy with recording density well below start of the art.
A high quality FM tuner is not only limited by the things you mention. If you have ever designed one, you might know.
You are not going to build an analogue chain that does better then those cheap digital players for that price, no way no how.
A $150 retail machine has a price from the importer/factory of no more then $75 or so, which means that the BOM cost cannot be greater then about $25 if the factory is to remain in business.
If 30% of that goes on the metalwork, plastic, packing and the box it comes in, 30% goes on the transport, that leaves you somewhere around $10 for non transport related electronics. A cheap DAC or codec chip, some generic R's &C's and a 5532 or so I can see for that, leaving money for a simple power supply, I am not seeing an fm discriminator having anything like the same performance for that BOM cost.
Besides, your 99% use low bitrate mp3s on poor quality earbuds.
Pretty much all modern digital audio production gets the dither to 16 bits right these days, that is a long solved problem.
Regards, Dan.
OK, we've made some progress. You're now in agreement that Red Book standard can have excellent performance, which was the original point. For the new position of your goalposts, you're more concerned with how cheap stuff performs. And I don't really have any data on that- I don't have a CD player, cheap or otherwise, and Stereophile doesn't measure them.
Since you're now also moving the goalposts to how recordings are engineered, can you give an example of a CD made after 1985 that has quantization distortion? I'd have no problem giving examples of badly mastered or recorded LPs or tapes.
Every Redbook level standard recording made whenever has noticeable quantization distortion in my setup. In every case where I have acquired a higher resolution version of the recording (such as 24 bit or decent analog), it has sounded considerably better & cleaner with more accurate detail. 48 Khz sampling sounds better to me than 44.1 khz, 96 sounds better yet. Only the lowest is available on CD.
I have dozens of CDs - not a single one strikes me as very accurate to the source material from what I could determine. None have accurate low level audio presentation. Not one images very well or consistently - certainly not outside the speakers. Many are grating to my ear. I become uncomfortable listening to most of them after a short period of time because of these qualities.
I listen to a few CDs repeatedly - most I do not.
I have dozens of CDs - not a single one strikes me as very accurate to the source material from what I could determine. None have accurate low level audio presentation. Not one images very well or consistently - certainly not outside the speakers. Many are grating to my ear. I become uncomfortable listening to most of them after a short period of time because of these qualities.
I listen to a few CDs repeatedly - most I do not.
Last edited:
Can you name just one CD with quantization distortion and how you determined this?
I've come across only one CD where there was a noticeable flaw - a classical, operatic excerpts from the mid 80's, by Kiri Te Kanawa, CBS. The first track has been mastered at a very low volume, absurdly so - maximum volume is barely loud enough ... and the flaw is due to some sloppiness, incompetence by the recording engineer, the extremely quiet start to the track has audible quantisation artifacts - looking in a waveform editor one can see the staircase noise in the signal ...
I've been reading this thread intently. Fascinating...
My 2 cents:
First Cent: I have an job on my college campus working sound, lighting, and video projection for events. I am the audio guy mostly. We use class D amps, obviously for the efficiency, but all of our signal processing and wireless mic systems are analogue because there is no encoding- the audio out of the speakers matches the person talking. I have seen digital systems have slightly noticeable 'delay'. key word is slightly. But analogue is instant for all intensive purposes.
Second Cent: Vinyl and tape has the potential to present more accurate detail and be more true to the original recording done in analogue in the studio, since it is analogue, and theoretically has infinite variability. I use the word potential for a reason though. A turntable 'plays' the vinyl- it is an instrument. The vinyl does not have data etched on it- it has sound etched onto it. The turntable plays this sound and turns it into an electrical signal to be amplified. CD players process digital data. Tape decks process analogue data. Vinyl plays sound.
And inherent to that is an issue with vinyl. You have to have a good instrument for the best sound. A good turntable is a very precisely tuned and optimized instrument to produce a very specific desired sound: the sound as it is on the vinyl and nothing more or less. But because of the way vinyl is played, there is a finite practical limit to how much detail can be extracted. The needle can only be so small, and the groove etched only so precisely. So there is a practical limit to the detail and accuracy that can be extracted. This applies to tape as well, since there is a finite practical limit to the reader head's accuracy and the accuracy of the signal magnetically traced to the tape. Tape decks also can be affected by variations in tape speed, head height, vibration, stray magnetic fields... all of which corrupt the audio signal. So tape decks too also require much design work and optimization.
Digital obviously has a limited resolution as well. Because it has to be transformed from digital to analogue to be listened to, the original signal has to contain a lot more data to be processed into a analogue waveform- a process that is not perfect and analogue sources need not worry about. But because it is digital, all the data is there. Upgrading equipment does not increase the ability for the source to pick up the data off of a CD or hard drive or process the data. The upgrade really relies in the conversion to analogue. Digital is digital. On or off. As long as it can be processed as on or off the ability for a digital source to posses detail that can be processed is independent of what is being used to read that source. The difference from low quality equipment to high quality equipment mostly (not entirely) lies in how accurately that data can be used to produce the analogue electrical signal.
So what does that come down to? There is more detail and lifelikness intrinsic to analogue sources. But with something analogue, the player of that analogue source has to be much more precise and fine tuned than the processor of a digital audio source. This makes high end analogue audio more expensive and more of a serious endeavor than digital music, and out of the reach of some. High end digital music does not rely on the source (assuming decent digital files) All that digital audio hinges on is a good DAC, which involves analogue. The Achilles Heel of decent sound from a digital source is the analogue component.
I have to say though, sound quality aside (because my turntable is cruddy and it is not set up well in my college dorm, and I listen to nothing but 16bit or higher FLAC files with an outboard DAC from my computer, so my 'digital' listening offers much much better sound) I prefer vinyl. Just because it's vinyl. With my current situation, I never do serious listening off of vinyl. But there is something about a turntable that transcends the music.
My 2 cents:
First Cent: I have an job on my college campus working sound, lighting, and video projection for events. I am the audio guy mostly. We use class D amps, obviously for the efficiency, but all of our signal processing and wireless mic systems are analogue because there is no encoding- the audio out of the speakers matches the person talking. I have seen digital systems have slightly noticeable 'delay'. key word is slightly. But analogue is instant for all intensive purposes.
Second Cent: Vinyl and tape has the potential to present more accurate detail and be more true to the original recording done in analogue in the studio, since it is analogue, and theoretically has infinite variability. I use the word potential for a reason though. A turntable 'plays' the vinyl- it is an instrument. The vinyl does not have data etched on it- it has sound etched onto it. The turntable plays this sound and turns it into an electrical signal to be amplified. CD players process digital data. Tape decks process analogue data. Vinyl plays sound.
And inherent to that is an issue with vinyl. You have to have a good instrument for the best sound. A good turntable is a very precisely tuned and optimized instrument to produce a very specific desired sound: the sound as it is on the vinyl and nothing more or less. But because of the way vinyl is played, there is a finite practical limit to how much detail can be extracted. The needle can only be so small, and the groove etched only so precisely. So there is a practical limit to the detail and accuracy that can be extracted. This applies to tape as well, since there is a finite practical limit to the reader head's accuracy and the accuracy of the signal magnetically traced to the tape. Tape decks also can be affected by variations in tape speed, head height, vibration, stray magnetic fields... all of which corrupt the audio signal. So tape decks too also require much design work and optimization.
Digital obviously has a limited resolution as well. Because it has to be transformed from digital to analogue to be listened to, the original signal has to contain a lot more data to be processed into a analogue waveform- a process that is not perfect and analogue sources need not worry about. But because it is digital, all the data is there. Upgrading equipment does not increase the ability for the source to pick up the data off of a CD or hard drive or process the data. The upgrade really relies in the conversion to analogue. Digital is digital. On or off. As long as it can be processed as on or off the ability for a digital source to posses detail that can be processed is independent of what is being used to read that source. The difference from low quality equipment to high quality equipment mostly (not entirely) lies in how accurately that data can be used to produce the analogue electrical signal.
So what does that come down to? There is more detail and lifelikness intrinsic to analogue sources. But with something analogue, the player of that analogue source has to be much more precise and fine tuned than the processor of a digital audio source. This makes high end analogue audio more expensive and more of a serious endeavor than digital music, and out of the reach of some. High end digital music does not rely on the source (assuming decent digital files) All that digital audio hinges on is a good DAC, which involves analogue. The Achilles Heel of decent sound from a digital source is the analogue component.
I have to say though, sound quality aside (because my turntable is cruddy and it is not set up well in my college dorm, and I listen to nothing but 16bit or higher FLAC files with an outboard DAC from my computer, so my 'digital' listening offers much much better sound) I prefer vinyl. Just because it's vinyl. With my current situation, I never do serious listening off of vinyl. But there is something about a turntable that transcends the music.
Some of you may want to check out the tape project:
the Tape Project
Also, I seem to recall an article somewhere that detailed the custom tape machine made by Keith Johnson of Reference Recordings. For that machine, I think he used an HF oscillator (much higher frequency than normal) to bias the tape.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.