I want to thank Thorsten and others for this enlightnining discussion on digital, and WHY it sounds that way. While I have chosen in my limited lifetime to specialize in analog design, it is both interesting and necessary to find out what has both improved and what has been 'denied' as being wrong with digital. The debate between Thorsten and PMA being extra enlightening to me. This debate has been with us for decades, and I am continuously put in a position to monitor (with relatively fresh ears) each year, the state of the art in digital recording and playback, including SACD. Sometimes it is pretty good, but I will stick to analog, myself.
Any relevance (even remote) with the components/construction vibration “acoustic signature” topic?
Technology for obtaining electromagnetic energy from sound waves - DigInfo TV - Tech News Videos From Japan | The latest technology, products, gadgets and scientific research direct from Tokyo
Regards
George
Technology for obtaining electromagnetic energy from sound waves - DigInfo TV - Tech News Videos From Japan | The latest technology, products, gadgets and scientific research direct from Tokyo
Regards
George
Very little has been said as to why anything sounds the way it does. SY played some live music recordings through a $29 USB to SPDIF dongle (not even in an enclosure) for me last night. Why there is still a business in $12,000 DAC's, etc. is beyond me.
It is almost impossible to make steep analog filter (to prevent aliases) with low ripple of amplitude response and same time with suitable phase response in the passband. First A/Ds had low sampling frequency (no oversampling, no sigma-delta 1 bit) and thus needed steep anti-alias filters. Huge box with poor response.
http://forum.blu-ray.com/audio-theo...standing-d-analog-digital-conversion-pcm.html
Last edited:
Pavel,
With respect, I suspect you do not know the story of the drunk guy, the police officer and the car keys...
I understand you are looking where the light is good, but the car keys are over there, where things are quite dark and murky.
Lest you you use a method that is appropriate to look for these "mythical" problems, you will not be able to measure them any better than for example PIM and TIM using standard THD+N.
But as I percieve that you very much DO NOT want to measure them, please do not change your measurement methods, for if you do, you may dislike the results, I sure do dislike them...
Ciao T
With respect, I suspect you do not know the story of the drunk guy, the police officer and the car keys...
I understand you are looking where the light is good, but the car keys are over there, where things are quite dark and murky.
Lest you you use a method that is appropriate to look for these "mythical" problems, you will not be able to measure them any better than for example PIM and TIM using standard THD+N.
But as I percieve that you very much DO NOT want to measure them, please do not change your measurement methods, for if you do, you may dislike the results, I sure do dislike them...
Ciao T
John,
I come form a hardcore analogue background (Industrial/Military process control in the 80's in eastern europe), but even then mixed signal started to show up. I got into Audio almost by accident (basically being blacklisted by the communists government for political reasons) even though my early forrays into electronics in my pre-teens where very much music/audio related.
For a long time you could have coloured me as "militant vinyly", to the point of complete dismissal of CD as Amusical. However, it bugged me and I kept pecking away at thing.
In reality I am much better at analogue and RF design than at digital design...
Hey, hold it, outside some conceptual prisons even digital signals are actually analogue and RF...
Oh well...
Ciao T
I come form a hardcore analogue background (Industrial/Military process control in the 80's in eastern europe), but even then mixed signal started to show up. I got into Audio almost by accident (basically being blacklisted by the communists government for political reasons) even though my early forrays into electronics in my pre-teens where very much music/audio related.
For a long time you could have coloured me as "militant vinyly", to the point of complete dismissal of CD as Amusical. However, it bugged me and I kept pecking away at thing.
In reality I am much better at analogue and RF design than at digital design...
Hey, hold it, outside some conceptual prisons even digital signals are actually analogue and RF...
Oh well...
Ciao T
Again, an example of your PIM measurement would help more than quite empty sentences
Please distinguish between THD+N (number, or plot of number vs. frequency) and FFT spectral analysis. FFT will show you jitter, e.g.
I am looking forward your PIM measurements.
TIM is very easy to measure, but no properly designed circuit has any measurable TIM. Just IM. TIM is a story of the past, .... "decades ago"...
Please distinguish between THD+N (number, or plot of number vs. frequency) and FFT spectral analysis. FFT will show you jitter, e.g.
I am looking forward your PIM measurements.
TIM is very easy to measure, but no properly designed circuit has any measurable TIM. Just IM. TIM is a story of the past, .... "decades ago"...
Last edited:
Scott,
A lot has been said, but arguably, it requires the ability to integrate this information with prior information, rather than outright rejection.
Well, given that most current Speakers do not give good sound when connected directly to a digital SPDIF signal the reason may be by far less beyond you than you think.
Of course, such truth are neither convenient nor orthodox, so it may be wiser to take the road of the ostrich, that noblest of all fowl, famous for it's way most effectvive and expedient way of dealing with things out of the ordinary (sticking it's head into the sand)....
Ciao T
A lot has been said, but arguably, it requires the ability to integrate this information with prior information, rather than outright rejection.
Well, given that most current Speakers do not give good sound when connected directly to a digital SPDIF signal the reason may be by far less beyond you than you think.
Of course, such truth are neither convenient nor orthodox, so it may be wiser to take the road of the ostrich, that noblest of all fowl, famous for it's way most effectvive and expedient way of dealing with things out of the ordinary (sticking it's head into the sand)....
Ciao T
Pavel,
As English is not your first (or even second) language, it may have escaped your notice that I used THD & N vs PIM/TIM as an analogy to point out how your FFT's where fundamentally incapable of showing things that are random or at least semi-random in nature (that is, the FFT will not resolve them to a single spectral line), that happen dynamically and affect only a single sample, any more than a 1KHz THD & N will reveal the presence of TIM and/or PIM.
For example, using 8K samples will suppress such a "fuzzy distortion" by around 40dB!
However, as I said, you have your deep seated prejudices and ayou are not interested in what really happens, so I see little point in debate.
Please continue as you do and pay no attention to the man behind the curtain...
Ciao T
Again, an example of your PIM measurement would help more than quite empty sentences
As English is not your first (or even second) language, it may have escaped your notice that I used THD & N vs PIM/TIM as an analogy to point out how your FFT's where fundamentally incapable of showing things that are random or at least semi-random in nature (that is, the FFT will not resolve them to a single spectral line), that happen dynamically and affect only a single sample, any more than a 1KHz THD & N will reveal the presence of TIM and/or PIM.
For example, using 8K samples will suppress such a "fuzzy distortion" by around 40dB!
However, as I said, you have your deep seated prejudices and ayou are not interested in what really happens, so I see little point in debate.
Please continue as you do and pay no attention to the man behind the curtain...
Ciao T
Hi,
Yet Peter Baxandall declared this particular kind of boxes to be "essentially sonically transparent" (that is, he was unable to distinguish them on insertion).
I have my doubt that he (who has sadly or perhaps thankfully left this plane of eXistenZ) would feel similar about a chain made with current day "24 Bit" ADC's and DAC's using anywhere between several percent to nearly 100% "fuzzy distortion" as fundamental operating principle...
Ciao T
Yet Peter Baxandall declared this particular kind of boxes to be "essentially sonically transparent" (that is, he was unable to distinguish them on insertion).
I have my doubt that he (who has sadly or perhaps thankfully left this plane of eXistenZ) would feel similar about a chain made with current day "24 Bit" ADC's and DAC's using anywhere between several percent to nearly 100% "fuzzy distortion" as fundamental operating principle...
Ciao T
Gee, Scott, you replied without answering my question, and then dropped a name that I don't know, either. Who is this Bob Adam?
If you want explanations, browse the AES library for John Vanderkooy's papers. But I think you're right to be concerned about the impulse response of SACD, if that's what you mean by instantaneous performance.
DSD is not so hard to make sense of via Google. Try DA, though. I have no idea what you guys are talking about, and Google really can't help. See Wikipedia for a funny list of possibilities, none of which are audio related.
Hi,
DA = Dielectric Absorbtion - wikipeda has an introduction under capacitors...
Ciao T
DA = Dielectric Absorbtion - wikipeda has an introduction under capacitors...
Ciao T
the full complex Fourier Transform of a time series is an exact Dual == contains all of the information from the time series - including the variation of each and every sample from noise, distortion, from any source
whether it makes certain phenomena more or less visible to the human eye/brain is the whole point of using both views: time and frequency domain
if the FFT isn't the "best" tool that doesn't mean you can get off with hand waving assertions - describe the signal, noise, distortion properties - in the time domain if that is easier
but without a definite technical description of the problem(s), measurable consequences you are simply not participating in a technical debate - just baiting
whether it makes certain phenomena more or less visible to the human eye/brain is the whole point of using both views: time and frequency domain
if the FFT isn't the "best" tool that doesn't mean you can get off with hand waving assertions - describe the signal, noise, distortion properties - in the time domain if that is easier
but without a definite technical description of the problem(s), measurable consequences you are simply not participating in a technical debate - just baiting
Thanks. Rather than complain that Wikipedia had no link from "DA" to the appropriate sections under Capacitor and Permittivity, I just added one. Maybe the next guy will not have to interrupt the conversation...
You need to consider the case of a discrete time, discrete level Fourier transform, not the mathematical model. The limits of sampling rate and accuracy will affect your results.
Many here have found that 16 bit recordings at 44,100 Hz are not as good as 24 bits at 192,000 hz. for reasons open to debate. So if you want to make a measurement of the difference, you will require a minimum of 25 bits at 384,000 hz. for your analysis. (Ignoring such issues as clock jitter, noise and other sample errors.) To be safe you might want to go to 26 bits at 768,000 Hz. But of course the equipment to do that doesn't exist.
- Status
- Not open for further replies.