TurnTables Sound better than Digital !!! - Really ???

[Saket]

Whoa...this discussion is now really going over our heads with all these mathematics & graphs & waves & what not...
I would like to point out a very simple thing. We listen to music for pleasure & relaxation...so why to deviate to such complexities? At times, I still remember that I used to have a small radio near my pillow with only AM/SW switching while sleeping on my terrace in summer nights. It was my summer vacation some 14-15 years back. I enjoyed listening to that simple radio so much that I couldnt sleep until channels shut down. Comparing that simple radio on these waves & graphs would now be a 'crime' but still it gave so much pleasure. The point is that what makes an individual connect to music? Be it a gramohone, turntable, radio, CDP, ipod or whatever..if an individual connects to that medium & enjoys it, it is what it should be for him. Thats what is called individual taste & preferences.
Like even if someone comes straight down from the heaven & tells me that mp3/ipod or any other digital media is the best & God has told him this fact, then also I would never give up my vinyl habbit to which I have connected since childhood.
I am not against digital & I enjoy my digital collection at home & in my car while travelling but when I am in a serious listening mode, I would spin those vinyls preferetially over same tracks on CD because I connect to the medium & enjoy it by thinking about it, looking at the big album art of yesteryears and all that. It gives me great satisfaction to look at the LPs spinning to the greatest accuracy on the strobe lamp and the only diamond I have (excluding my wife) rubbing on vinyls to produce the thump on my speakers.

For some, the tea may be the same but in a different cup!

Regards,
Saket

 

[reignofchaos]

Here is a better experiment. You bring vinyl recording of a song and I bring a CD containing the same song.

First, you calculate the FT of the track on the LP. Then I calculate the DFT of the track on the CD.

After we are done with the last step, you calculate the inverse FT of your output and I do the inverse DFT of my output.

Then we compare whose result is closer to the original.

Game?

Whats the point of this? Do you think a record player does fourier transforms? :lol:

Arguing for the sake of just arguing is not cool!

 

[jls001]

Do people actually perform Discrete Fourier (or even Cosine) Transforms operations at home?

 

[Thad E Ginathom]

Well, I would hope that they don't do it in public! :lol:

The graphs are wrongly depicted.
The first graph ONLY shows the sampled data points.
It is absolutely WRONG to assume that this is what you will hear. (and hence draw a conclusion that you will not hear a pure sine wave)

Relieved that you have pointed this out. it just shows how data can be misused and misrepresented.

Anyone can zoom in to sample level in an audio editor, point to the line of dots and say, look how far apart they are, what's in the middle? Or point to the jagged line and say how can this be music? It is exactly the same kind of data misuse that hifi manufacturers do when their marketing depts publish performance graphs that are not outright lies, but show only the complimentary part of the truth.

 

[reignofchaos]

Do people actually perform Discrete Fourier (or even Cosine) Transforms operations at home?

Your dac does quite a few while playing music :lol:

 

[jls001]

Your dac does quite a few while playing music :lol:

Yeah, I understand the part about lots of processing and filtering happening in the DCT, FFT, etc domains within the DAC chips or upsampler chips.

But do folks (or at least some folks) have hardware FPGAs or DAC chips or generic DSP chips on which they perform such sophisticated digital signal processing, at HOME?

Or may be some sophisticated Mathlab-type application?

Let me make it clear that I am in awe here if there are such folks in our midst.

 

[Thad E Ginathom]

Here is a better experiment. You bring vinyl recording of a song and I bring a CD containing the same song.?

No: for the umpteenth time: you bring the vinyl, I'll digitise --- and you tell me which copy is being played. That, and only that, is the game.

Music companies do not put the same stuff on CD they put on vinyl.

For starters, they do not have to distort the music unrecognisably to record it onto CD :clapping:

This post is a troll. Yes, I know --- but its all good fun. And it's true

 

[thatguy]

Whats the point of this? Do you think a record player does fourier transforms?[/QUOTE]

No I don't. What gave you the idea? Besides, neither does a DAC have to do Fourier transforms.

Of all the problems with digital audio, you chose to criticise it for having a capability that that the analog world doesn't have. The last time I studied signal processing, you couldn't even do a Fourier transform of an arbitrary analog signal without first sampling it. It is really upto the audio engineer to decide whether he/she wants to do DFT/iDFT or not.

Arguing for the sake of just arguing is not cool!

Sorry, too old for cool/not-cool game. Stay with the facts, avoid spreading FUD and we won't need these 'pointless' arguments.

 

[reignofchaos]

Yeah, I understand the part about lots of processing and filtering happening in the DCT, FFT, etc domains within the DAC chips or upsampler chips.

But do folks (or at least some folks) have hardware FPGAs or DAC chips or generic DSP chips on which they perform such sophisticated digital signal processing, at HOME?

Or may be some sophisticated Mathlab-type application?

Let me make it clear that I am in awe here if there are such folks in our midst.

Its pretty easy to setup a simple library (brute force) for DFT/iDFT or DCT/iDCT in C. Doesn't take more than an hour of coding. Its like the foundation for any DSP or Image processing course. However optimizing it for computational performance is a research topic in itself.

In matlab/octave or a similar app, a DFT/iDFT on a vector/image is a single operation. There's not much code writing involved . In fact you can do it on a real world audio file (.wav) without much extra effort.

 

[reignofchaos]

Whats the point of this? Do you think a record player does fourier transforms?[

No I don't. What gave you the idea? Besides, neither does a DAC have to do Fourier transforms.

Any dac that has an upsampler does operations in the frequency domain - which basically means almost every good dac. How would you get your signal to the frequency domain without a DFT or a DCT?

Of all the problems with digital audio, you chose to criticise it for having a capability that that the analog world doesn't have. The last time I studied signal processing, you couldn't even do a Fourier transform of an arbitrary analog signal without first sampling it. It is really upto the audio engineer to decide whether he/she wants to do DFT/iDFT or not.

There you go you answered yourself. Sampling itself is lossy cos you obviously can't store the infinite number of coefficients that are needed for reconstructing the analog signal exactly. What does that mean? Digital audio is an approximation right from the word go - whether it is better or worse than a particular analog system depends on implementation of the system. It is purely an engineering problem.

Sorry, too old for cool/not-cool game. Stay with the facts, avoid spreading FUD and we won't need these 'pointless' arguments.

Unfortunately you are the one who is spreading FUD not me.

 

[thatguy]

What does that mean? Digital audio is an approximation right from the word go

Well, even the grooves in the vinyl are an approximation of the original sound wave created by the musical instrument or the singer's voice. The trajectory of the needle is a further approximation of the groove. The real question is: after all the processing is done, which one's output closer to the original. Unless you bring in numbers which can be compared, it is really one opinion vs another. Examining one idea like the DFT (or dithering or sampling) under a microscope is missing the big picture.

Anyway, the professionals who *really* know this stuff and make a living producing music have made their choice(except the few who are still holding out), in spite of the DFT and the iDFTs. Not sure if there is a better argument for digital audio. Statements like 'Sampling itself is lossy cos ...' sound hollow in this day.

Upsampling is interpolation. Why do you need to do a DFT? Filters maybe, but DFT?

 

[ynotbubba]

Cool down guys... :lol:

Here are some refreshments:

Digital vs Analog

 

[KiranPS]

Guys,

Sorry for my Ignorance when I posted the Images about the sampled waveforms.

Can someone please tell me if a DAC in a CD Player generates the Square wavish output like the image from the PCM data or does it do Bezier type curve fitting (or some other Fourier TFM may be with some DSP.) to generate actual sine wave ? Is this why the CD players from marantz/onkyo are more expensive than the consumer class Philips ones?

http://www.hifivision.com/attachmen...sound-better-than-digital-really-14khz-44.gif

 

[reignofchaos]

Anyway, the professionals who *really* know this stuff and make a living producing music have made their choice(except the few who are still holding out), in spite of the DFT and the iDFTs. Not sure if there is a better argument for digital audio. Statements like 'Sampling itself is lossy cos ...' sound hollow in this day.

What is so hollow about the truth? Digital music an approximation. The only thing that is hollow here is the ranting that is going on. Digital audio playback is an engineering problem and so there will always be improvements.

Upsampling is interpolation. Why do you need to do a DFT? Filters maybe, but DFT?

If you've ever looked at the code of any real time upsampler, you'd not be asking me that question. Any sample rate converter has a low pass filter after the sinc function stage to avoid spuriae. Also many hardware chips actually do the convolution with the sinc in frequency domain to reduce complexity.

Need software based examples? Look up the code of libsamplerate - it is public domain.

 

[thatguy]

Digital music an approximation.

Why are you using the word approximate with only digital. Both vinyl and digital are approximate. How about getting some real numbers about the distortion+noise of the analog chain (analog master -> vinyl -> phono out) vs digital chain (analog masters -> CD/SACD -> DAC out). Making claims about a DFT/iDFT destroying your music without specifying the quantity of the change is pointless.

Digital audio playback is an engineering problem and so there will always be improvements.

It is not just an engineering problem. It is backed by pretty solid theory.

If you've ever looked at the code of any real time upsampler, you'd not be asking me that question. Any sample rate converter has a low pass filter after the sinc function stage to avoid spuriae. Also many hardware chips actually do the convolution with the sinc in frequency domain to reduce complexity.

Need software based examples? Look up the code of libsamplerate - it is public domain.

Thanks for the info. I will take a look at libsamplerate. Still unsure about the DFT part, but I need to read up.

 

[Thad E Ginathom]

Why are you using the word approximate with only digital. Both vinyl and digital are approximate.

Very balanced. The act of recording and playback is an approximation. Engineers at both ends of the chain can improve what we have and even come up with better, different, technologies.

 

[jls001]

Why are you using the word approximate with only digital. Both analog and digital are approximate. How about getting some real numbers about the distortion+noise of the analog chain (analog master -> vinyl -> phono out) vs digital chain (analog/digital masters -> CD/SACD -> DAC out). It is fine if you can't get it. But then talking about superiority of one one medium over the other doesn't mean much.

Everything is an approximation of the real thing.

For example a singer sings into a microphone in a studio voice booth, the mic adds or subtracts something from the original signal, as does the cable carrying the mic signal to the mixing console, console adds/subtracts further, the recording device may not be able to record the complete event that reaches it even if it is professional analog tape (tape wow, hysteresis effect, etc) or even hard disk based digital recoding (which samples at some frequency and stores data at some bit depth).

The mixing engineer does his own thing and as does the final mastering engineer.

What we get to play in our homes is again subject to how well our audio chains can reproduce the mastered program material.

You are right when you say that vinyl suffers from much inferior measurements (S/N ratio in the 70s, channel separation in the mid 20s, eroded groove causing surface noise, dust particles causing clicks and pops, variability of playback system, etc - I'm sure there are more in the 'analog problem set').

But digital CDs also have its share of issues despite having an enviable S/N ratio hovering near 100 dB.

But to come back to point being contested here, digital IS an approximation by virtue of the non-feasibility of having infinite samples to reconstruct the original signal. Real-world system - electrical, electronic, mechanical - are engineering compromises and are not designed for handling something infinite.

My point is not to reinforce the superiority of one medium over the other.

I suggest everyone should try to listen to a properly set up analog system as well as a well-thought out digital rig.

I'm told that at the highest level, both types of system can be beguilingly similar. And beautiful.

 

[thatguy]

But to come back to point being contested here, digital IS an approximation by virtue of the non-feasibility of having infinite samples to reconstruct the original signal. Real-world system - electrical, electronic, mechanical - are engineering compromises and are not designed for handling something infinite.

Cross posting:

Very balanced. The act of recording and playback is an approximation. Engineers at both ends of the chain can improve what we have and even come up with better, different, technologies.

I am not going to contest that digital (even in theory) is an approximation. Primarily because it doesn't satisfy the criteria of the sampling theorem. Sampling theorem only holds for band limited signals. One wouldn't know where to put the anti-aliazing/imaging filters if the signal was not band limited. A 5 minute song is time limited => not band limited. (as a general rule, a time limited signal can't be band limited and a band limited signal can't be time limited).

However, when we use the word approximate we should some idea of 'how much'. This is in the domain of mathematics. Are the artifacts introduced by sampling at -30dB or -90dB or -150dB? Is it comparable to the degradation that a magnetic tape suffers with time or a vinyl suffers with each playback? Is it possible to reduce the artifacts to below audible levels by increasing the sampling rate/depth?

I am quite certain a lot of scientists and engineers have wrecked their head over these problems and many of the answers exist, even when *we* don't know about them. A breezy declaration of victory by focusing on one minute aspect of a process baffles me.

 

[captrajesh]

Here's an explanation of Class A, B, A/B, etc: http://www.hifivision.com/amplifiers/174-types-amplifiers-class-class-b-class-ab-class-d.html

I wanted to reply to Gerry_the Merry's post and was searching for this very thread which I had subscribed to and searched in vain till 2:30 A.M yesterday. The upside is, I was able to unsubscribe to about 50 odd threads which I no longer wish to participate in.

Captain - I don't know if you disagree with me or if you are asking for an explanation for your benefit.

With due respect to your knowledge and understanding, I beg to differ with you on some issues. I'll take the points that I disagree one by one.

class AB amplifier has at very low output, class A operations, but at higher volumes, switches to class B/AB.

AFAIK, not all Class AB amps are designed to run in Class A mode at low outputs. As per my understanding, the difference between Class A and Class AB are like this. In class A amps, a constant current as per the bias point set continues to flow through the out output device (transistor/Tube) whereas in Class AB amp, the bias point is set in such a way that no current flows through the output device at idle. Some Class AB amps are so biased that a small amount of current keeps flowing even at idle. Only such amps are capable of operating in Class A topology for the first few watts which corresponds to the bias current.

The lossy-ness pertains to the fact that unlike in class A, where in a sine wave component, the entire 360 degrees is processed off a linear profile from input to output, in a class B operation, only 180 degrees is processed, and this creates a distortion at 0 degrees. To overcome this, a mirror image signal is generated and integrated at the output stage to synthetically create a full 360 degrees signal, but since the output function kicks in only at some minimum voltage (like 0.5mV, if my understanding is correct)

I think you are describing the mode of operation of a Class B amp where the top and bottom halves of the sine wave are amplified by two discrete output devices and the amplified and reconstructed sine wave carries intermodulation distortion which is quite audible. This limitation has been largely overcome by using the Class AB topology where a little more than 180 degrees of waveform is amplified and joined together.

there is a loss between 0 and +- 0.5V. This loss cannot be compensated. however at higher frequencies, this distortion becomes smaller and smaller as the cutoff voltage is reached quicker and quicker. But at lower frequencies, there is a definite irretrievable loss. I don't know if I have succeeded in explaining to you, but folks who are familiar with this subject will perhaps be able to explain better than I have managed.

I'm not sure of what is stated here. I found the explanation of the Class A and AB provided at the site linked below very easy to understand. May be you would find it interesting too.

Class A - Exposed and Explained by Randall Smith

 

[ranjeetrain]

It's surprising when analog fans go out to bash digital for the "lost data", which has been considered beyond the human hearing limit by scientists and researchers (probably all of them 100s or 1000s folds more knowledgeable on the subject than everyone participating in this thread), while ignoring the fact that analog medium suffers with even more problems.

The fact also has it that "the loss" in digital medium is only a ONCE phenomena. After a signal has been approximated, it never changes on the storage medium. Since these losses are predictable they can be attempted to be reconstructed in an educated manner.

On the other hand, losses in analog medium are non-stop. From mastering to pressing there is a loss. Has anyone done any research on what %age of original signal is present on a pressed vinyl soon after it has been produced? Isn't the pressing machine susceptible to "pressing errors". Are those losses less significant than the "approximation loss"? Is there any measurement around this? If analog fans are so worried about "approximation errors", they should not ignore the "pressing errors". These errors may as well be of a higher magnitude than the one that worries them about digital.

Losses in analog medium continue to occur.

In typical Indian households dust is a major problem. What is the size of grooves on an LP? What is the size of micro particles floating in the error, that settle down on the disc while they are being used? Does that not change the "original signal"?

Analog fans conveniently ignore all the problems associated with analog while they are busy declaring digital imperfect due to "approximation errors".

Finally, one question for everyone to think about. Once a track has been mastered and stored digitally, two copies of the same track will sound absolutely identical even after decades of use and 1000s of playback. Can the same said to be true for analog? Two LPs containing same track, played on different systems by different people a different number of times, will they sound identical?

My conclusion is that: digital suffers from a single (and predictable) flaw (loss in rounding/approximation), but analog suffers from number of flaws. A track mastered digitally in a good studio at a high sampling rate will any day beat analog, for all intent and purposes.