IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Higher Sample Rate When Capturing From Slower Analog Playback Source
JasonCA
post Jan 6 2013, 22:16
Post #1





Group: Members
Posts: 4
Joined: 6-January 13
Member No.: 105649



Hi Everyone,

I'm wondering if there is a loss of audio quality by capturing from a recorded analog audio source at a slower playback rate? This applies to those who may not have a device that plays back at a higher playback rate, but instead can only playback at a slower playback rate.

I've seen some people who have had no choice but to playback vinyls on say at 45 RPM since those machines can't playback at 78 RPM. They can then pitch shift the audio digitally to get the playback speed correct. But, the question is...is there audio loss? At a slower playback speed, my understand is you get a higher sample rate on your audio source. With a higher sample rate, wouldn't the quality even better than having had captured it at the normal 78RPM's for a 78 vinyl record?

In this case, I'm not asking specifically for vinyl...but more or less in general. For example, let's look at an audio cassette tape deck as another example that plays back audio at half the normal playback speed. My understanding is that the if you were to record the audio when the cassette tape plays back at half the normal playback speed, you are essentially increasing the sample rate of the audio? True? Or? So when going back to the normal playback speed, you would essentially be digitally down-sampling your captured audio to create a final audio file in which so far to me wouldn't be any worse than to have captured it at the normal playback speed to begin with?

People have said that even though you save an audio source digitally at a down-sampled rate (ex: saving final audio at 44 kHz by down-sampling the captured 192 kHZ audio file), that internally the sound card is already capturing the data at the highest sample rate and then down-sampling it when it passes it off to the software when the software application is recording at a lower sample rate than the sound card is capable. So some would argue that there is no difference between a 192 kHz final audio source file over a 44 kHZ because the 44kHz file was internally created by the sound card having down-sampled the 192 kHZ audio to create it in the first place.

Going back to my original question, what I'm trying to understand is if there is a loss of audio quality by down-sampling or essentially pitch shifting audio that was captured at a slower playback rate? To me, it seems as if the audio quality would instead be better because by playing your analog source back a slower playback rate, you allow the sound card (internally or otherwise) to sample the audio at a higher sample rate than it would have had the chance to do if you played it back at a faster rate. So, to me it would seem the audio quality may even be 'better' by capturing the audio at a slower playback rate?

So whether the audio source is vinyl records or audio cossets or otherwise, does capturing audio data at a slower analog playback speed effect the final audio quality? And, if so why? Or, does it result in essentially the same? Or, is the audio quality even better?
Go to the top of the page
+Quote Post
saratoga
post Jan 6 2013, 22:33
Post #2





Group: Members
Posts: 5042
Joined: 2-September 02
Member No.: 3264



QUOTE (JasonCA @ Jan 6 2013, 16:16) *
I've seen some people who have had no choice but to playback vinyls on say at 45 RPM since those machines can't playback at 78 RPM. They can then pitch shift the audio digitally to get the playback speed correct. But, the question is...is there audio loss? At a slower playback speed, my understand is you get a higher sample rate on your audio source. With a higher sample rate, wouldn't the quality even better than having had captured it at the normal 78RPM's for a 78 vinyl record?


On modern A/D converter's, the clock rate of the actual conversion is more or less constant. If you double the sampling rate, the converter halves the oversampling ratio to maintain the clockspeed its designed for. So if you take twice as long, you effectively do get twice as many samples, and thus sqrt(2) better SNR.

However, as no analog capture device is limited by the SNR of a moderately good A/D, this gain is of negligible value in practical applications. Instead, I would be much more concerned about distortion resulting from playing back an analog device at other then it's design speed. Few mechanical systems are truly time invariant in the way that digital systems are.

Go to the top of the page
+Quote Post
JasonCA
post Jan 6 2013, 23:07
Post #3





Group: Members
Posts: 4
Joined: 6-January 13
Member No.: 105649



I appreciate your feedback! smile.gif

QUOTE (saratoga @ Jan 6 2013, 23:33) *
On modern A/D converter's, the clock rate of the actual conversion is more or less constant. If you double the sampling rate, the converter halves the oversampling ratio to maintain the clockspeed its designed for. So if you take twice as long, you effectively do get twice as many samples, and thus sqrt(2) better SNR.


Ok, so the idea is good. Your saying that if you playback the analog source at half the normal playback speed of the audio source, then you are effectively getting twice as many samples. And, you are saying the SNR is therefore improved. But...you are also more concerned about the caveat...

QUOTE (saratoga @ Jan 6 2013, 23:33) *
However, as no analog capture device is limited by the SNR of a moderately good A/D, this gain is of negligible value in practical applications. Instead, I would be much more concerned about distortion resulting from playing back an analog device at other then it's design speed. Few mechanical systems are truly time invariant in the way that digital systems are.


So you are saying here that halving the playback speed of the normal playback speed for a given along audio source, you are worried that the mechanical system or device is not capable of correctly reproducing an analog audio signal that it would normally be able to produce at a faster playback rate?

What about for devices that have a speed selector? There are tape decks (of various kinds) where you can play back the tapes at different speeds (user selectable way to change the IPS of playback)? So sometimes the device is capable of controlling playback. So what if the analog source was recorded originally at a normal playback speed, but then there is a half speed tape IPS switch? If the device supports halving the speed by design, wouldn't this imply that the device is capable of playing back the analog source mechanically at a slower rate?

In terms of vinyl record, I've read arguments that a vinyl record played back on a 45 RPM machine doesn't equate to the same audio output if the vinyl record was originally recorded at 78 RPM. The argument was the needle works better/different on a 78 RPM turntable then it does on a 45 RPM. How true or untrue this is, I'm not sure.

But, in regards to cassette tapes, I don't see why there would be mechanically anything different with a tape head reading analog tape back at slower playback rate especially if the device is capable of adjusting the playback rate of the tape?

So maybe there would be no audio degradation with audio cassettes, but there would be with vinyl records? It's just to say that for maybe some analog source devices, capturing at a slower playback rate would increase the sampling rate on capture and would result in a final digital capture that is no worse (but may be better) than if it were captured at the normal playback rate? Or, maybe there is no real way to know.
Go to the top of the page
+Quote Post
Soap
post Jan 6 2013, 23:28
Post #4





Group: Members
Posts: 1016
Joined: 19-November 06
Member No.: 37767



If I may be so bold:

He's saying that even if you have a mechanical playback system which does not degrade the quality when playing back at fractional speed you haven't gained anything since the A/D converters are already orders of magnitude better than the analog source. So why even consider the added headache and risk.

so:

QUOTE
It's just to say that for maybe some analog source devices, capturing at a slower playback rate would increase the sampling rate on capture and would result in a final digital capture that is no worse (but may be better) than if it were captured at the normal playback rate?


is exactly backwards.

The theoretical gains in playback at fractional speed is an increase in sampling SNR - a problem which does not exist as any competent ten cent A/D already far exceeds the source. Therefore the proper thought is "capturing at a slower playback rate would increase the sampling rate on capture and would result in a final digital capture that is no better (but may be worse)"

This post has been edited by Soap: Jan 6 2013, 23:31


--------------------
Creature of habit.
Go to the top of the page
+Quote Post
ktf
post Jan 6 2013, 23:43
Post #5





Group: Members
Posts: 394
Joined: 22-March 09
From: The Netherlands
Member No.: 68263



QUOTE (JasonCA @ Jan 6 2013, 23:07) *
But, in regards to cassette tapes, I don't see why there would be mechanically anything different with a tape head reading analog tape back at slower playback rate especially if the device is capable of adjusting the playback rate of the tape?

Read this: http://en.wikipedia.org/wiki/Helical_scan

Especially this:
QUOTE
In a fixed tape head system, magnetic tape is drawn past the head at a constant speed. The head creates a fluctuating magnetic field in response to the signal to be recorded, and the magnetic particles on the tape are forced to line up with the field at the head. As the tape moves away, the magnetic particles carry an imprint of the signal in their magnetic orientation. If the tape moves too slowly, a high frequency signal will not be imprinted: the particles' polarity will simply oscillate in the vicinity of the head, to be left in a random position. Thus the bandwidth channel capacity of the recorded signal can be seen to be related to tape speed: the faster the speed, the higher the frequency that can be recorded.

This is regarding recording at low speed, but it applies to playing as well. A tape head produces a signal because a tape is moved in front of it: the coil of the head moves relative to the fluctuating magnetic field of the tape, this is called induction. So, the movement of the tape is actually generating electricity. If you play the tape slower then usual to the extreme of not moving at all, there will be no signal. The faster the tape moves, the more signal it will generate.

So, slowing down the tape will actually add noise, because there is less voltage produced at the head.


--------------------
Music: sounds arranged such that they construct feelings.
Go to the top of the page
+Quote Post
JasonCA
post Jan 6 2013, 23:53
Post #6





Group: Members
Posts: 4
Joined: 6-January 13
Member No.: 105649



You can certainly be bold Soap. This is not a personal matter for me, I'm simply trying to come to a proper understanding. So be as bold as you like smile.gif. If I learn something, great!

QUOTE (Soap @ Jan 7 2013, 00:28) *
If I may be so bold:

He's saying that even if you have a mechanical playback system which does not degrade the quality when playing back at fractional speed you haven't gained anything since the A/D converters are already orders of magnitude better than the analog source. So why even consider the added headache and risk.


Why consider the risk? Some devices don't support full playback speed (or at least of ones I may own). Do I go crazy and go out and spend money for a device that plays at that speed? Or, can I use a device that plays half the speed, down-sample in software, and end up with the same result? Pick the bigger headache? If you have a device that plays at half the speed, why get a device that plays at the proper speed if all concerns can be addressed in software? Hence my question. So based on the questions that were raised, that is all I'm trying to understand: what are the risks? Do I loose audio quality? Or, am I maybe gaining audio quality at at best and not loosing it? Prior to my question, I (and perhaps others) could assume that maybe you gain by capturing at a slower playback rate. But if you don't, it's why I'm asking...to understand smile.gif


QUOTE (Soap @ Jan 7 2013, 00:28) *
so:

QUOTE
It's just to say that for maybe some analog source devices, capturing at a slower playback rate would increase the sampling rate on capture and would result in a final digital capture that is no worse (but may be better) than if it were captured at the normal playback rate?


is exactly backwards.

The theoretical gains in playback at fractional speed is an increase in sampling SNR - a problem which does not exist as any competent ten cent A/D already far exceeds the source. Therefore the proper thought is "capturing at a slower playback rate would increase the sampling rate on capture and would result in a final digital capture that is no better (but may be worse)"


This is the sticking point at the moment. I don't understand why it may be worse? If the sampling rate has increased, then I don't understand how you could get a worse outcome? To me it's like saying I have a source playing back at a 384 kHz sampling rate, but because I have such a high sampling rate, producing that 44 kHz final output file may result in worse quality because the 384 kHz has a higher sample rate. That would not make much sense to me. But, perhaps I'm confusing things! By all means, please correct my understanding! wink.gif

Go to the top of the page
+Quote Post
Soap
post Jan 6 2013, 23:57
Post #7





Group: Members
Posts: 1016
Joined: 19-November 06
Member No.: 37767



QUOTE (JasonCA @ Jan 6 2013, 17:53) *
You can certainly be bold Soap. This is not a personal matter for me, I'm simply trying to come to a proper understanding. So be as bold as you like smile.gif. If I learn something, great!

If I may be so bold as to speak for saratoga.

QUOTE (JasonCA @ Jan 6 2013, 17:53) *
Why consider the risk? Some devices don't support full playback speed (or at least of ones I may own).

Well, if you can't playback at full speed the issue is moot.


QUOTE (JasonCA @ Jan 6 2013, 17:53) *
This is the sticking point at the moment. I don't understand why it may be worse? If the sampling rate has increased, then I don't understand how you could get a worse outcome? To me it's like saying I have a source playing back at a 384 kHz sampling rate, but because I have such a high sampling rate, producing that 44 kHz final output file may result in worse quality because the 384 kHz has a higher sample rate. That would not make much sense to me. But, perhaps I'm confusing things! By all means, please correct my understanding! wink.gif

Where it may be worse is not in the digital side of things. You're looking in the wrong place.
Where it may be worse is that the analog playback equipment very likely does not perform the same at half speed as it does at full speed.
ktf just gave a very good example of this - though you may have been typing while he posted it.

This post has been edited by Soap: Jan 6 2013, 23:58


--------------------
Creature of habit.
Go to the top of the page
+Quote Post
saratoga
post Jan 7 2013, 00:00
Post #8





Group: Members
Posts: 5042
Joined: 2-September 02
Member No.: 3264



QUOTE (JasonCA @ Jan 6 2013, 17:53) *
Why consider the risk? Some devices don't support full playback speed (or at least of ones I may own). Do I go crazy and go out and spend money for a device that plays at that speed? Or, can I use a device that plays half the speed, down-sample in software, and end up with the same result?


I don't think any of us can answer this question without first taking the specific device you have and then seeing how well it works.


QUOTE (JasonCA @ Jan 6 2013, 17:53) *
This is the sticking point at the moment. I don't understand why it may be worse?


Well it won't better, and it would probably take some luck to get it to be the same. Hence I would assume it will be somewhat worse. Try it and see.
Go to the top of the page
+Quote Post
JasonCA
post Jan 7 2013, 00:33
Post #9





Group: Members
Posts: 4
Joined: 6-January 13
Member No.: 105649



QUOTE (Soap @ Jan 7 2013, 00:57) *
QUOTE (JasonCA @ Jan 6 2013, 17:53) *
You can certainly be bold Soap. This is not a personal matter for me, I'm simply trying to come to a proper understanding. So be as bold as you like smile.gif. If I learn something, great!

If I may be so bold as to speak for saratoga.

QUOTE (JasonCA @ Jan 6 2013, 17:53) *
Why consider the risk? Some devices don't support full playback speed (or at least of ones I may own).

Well, if you can't playback at full speed the issue is moot.


QUOTE (JasonCA @ Jan 6 2013, 17:53) *
This is the sticking point at the moment. I don't understand why it may be worse? If the sampling rate has increased, then I don't understand how you could get a worse outcome? To me it's like saying I have a source playing back at a 384 kHz sampling rate, but because I have such a high sampling rate, producing that 44 kHz final output file may result in worse quality because the 384 kHz has a higher sample rate. That would not make much sense to me. But, perhaps I'm confusing things! By all means, please correct my understanding! wink.gif

Where it may be worse is not in the digital side of things. You're looking in the wrong place.
Where it may be worse is that the analog playback equipment very likely does not perform the same at half speed as it does at full speed.
ktf just gave a very good example of this - though you may have been typing while he posted it.


Saratoga, SOAP, KTF did we just become best friends? Saying that reminds me of this Best Friends (not that I am at all crazy about the movie either...I'm not).

Awwwwww thanks KTF. So it seems it is therefore always best to playback the source at it's original recorded speed for the mechanical reasons stated by KTF. I would imagine the same sort of issue would also be the same for vinyl records and playing back at different speeds. Very interesting...I'll spend some more time to read and better understand the mechanical side a bit more. But, I think that sort of answers my main question. The fact that the audio signal can be degraded at a slower playback rate seems to me to be the pitfall.

Yes, I must have been writing when KTF was writing smile.gif.

Thanks for helping me to understand things a bit more and also helping to correct my understanding of how to better approach the original ideas I had. I'll go and ponder this a bit more now smile.gif
Go to the top of the page
+Quote Post
2Bdecided
post Jan 7 2013, 15:33
Post #10


ReplayGain developer


Group: Developer
Posts: 5175
Joined: 5-November 01
From: Yorkshire, UK
Member No.: 409



I think this has been discussed before. You need a way of capturing without RIAA, or compensating for RIAA - RIAA is wrong for 78s anyway.

Here you go...
http://www.hydrogenaudio.org/forums/index....=93659&st=0
...plenty of advice and discussion!

Equaliser from here will solve your RIAA EQ problems, whatever speed you record at...
http://www.clickrepair.net/software_downlo...nloads_win.html

Cheers,
David.
Go to the top of the page
+Quote Post
Arnold B. Kruege...
post Jan 7 2013, 16:07
Post #11





Group: Members
Posts: 3939
Joined: 29-October 08
From: USA, 48236
Member No.: 61311



QUOTE (JasonCA @ Jan 6 2013, 16:16) *
I'm wondering if there is a loss of audio quality by capturing from a recorded analog audio source at a slower playback rate?


As others have pointed out, running tape slow may cause loss of dynamic range.

However, dynamic range isn't anything like the whole picture. Bandwidth is also essential within reason.

The same argument may apply to vinyl because most vinyl transducers are also velocity sensitive.

QUOTE
I've seen some people who have had no choice but to playback vinyls on say at 45 RPM since those machines can't playback at 78 RPM. They can then pitch shift the audio digitally to get the playback speed correct.


They also have to adjust the eq. eq on 78s is always tricky because of the lack of a global standard.

QUOTE
But, the question is...is there audio loss? At a slower playback speed, my understand is you get a higher sample rate on your audio source. With a higher sample rate, wouldn't the quality even better than having had captured it at the normal 78RPM's for a 78 vinyl record?


See previous comments about dynamic range and bandwidth. Part of the craft of audio is trading off that which is less important sonically to get more of what is sonically more important. The trade offs can change for every different situation.

QUOTE
In this case, I'm not asking specifically for vinyl...but more or less in general.


All generalities are false including this one! ;-)

QUOTE
For example, let's look at an audio cassette tape deck as another example that plays back audio at half the normal playback speed. My understanding is that the if you were to record the audio when the cassette tape plays back at half the normal playback speed, you are essentially increasing the sample rate of the audio?


Sample rate is probably the wrong criteria and the wrong term. Good modern ADCs are so much better than any legacy analog media that they are not really part of this discussion.

Let's follow Shannon if we want to be generalized. It is all about bandwidth and dynamic range.

If you increase the tape speed you increase the voltage at the output of the tape head which potentially increases dynamic range, but then the width of the gap in the tape head and other frequency-dependent losses can come around bite you in the bandwidth.

If you increase the disk rotation speed you increase the voltage at the output of the cartridge which potentially increases dynamic range, but then the shape of the stylus and other frequency-dependent losses can come around bite you in the bandwidth.

If you decode the disk by means of imaging, then the resolution of your optical sensor is of the essence, and that has been a situation with most of such uses that I am aware of.

What do you need more of, dynamic range or bandwidth?

It is obvious to me that there is no perfectly general answer.

Every different case is different!

Study the situation, consider the costs and benefits, and try to get the best sounding results.

Go to the top of the page
+Quote Post
Ethan Winer
post Jan 7 2013, 21:35
Post #12





Group: Members
Posts: 248
Joined: 12-May 09
From: New Milford, CT
Member No.: 69730



QUOTE (JasonCA @ Jan 6 2013, 16:16) *
So whether the audio source is vinyl records or audio cossets or otherwise, does capturing audio data at a slower analog playback speed effect the final audio quality? And, if so why? Or, does it result in essentially the same? Or, is the audio quality even better?


Unless I missed it, I didn't see anyone comment on the common practice of using half-speed to improve high frequency response. When a record is played at a slower speed, the cartridge can more easily track high frequencies with less "spitting" type distortion. But yes, RIAA (records) and NAB (tape) curves will not be accurate.

Personally, if I had old records or tapes that I recorded digitally, I'd probably EQ them by ear to sound as I like. This could possibly improve on the original tonality, and would also compensate for improper RIAA and NAB playback response curves.

--Ethan


--------------------
I believe in Truth, Justice, and the Scientific Method
Go to the top of the page
+Quote Post
Arnold B. Kruege...
post Jan 8 2013, 15:43
Post #13





Group: Members
Posts: 3939
Joined: 29-October 08
From: USA, 48236
Member No.: 61311



QUOTE (Ethan Winer @ Jan 7 2013, 15:35) *
QUOTE (JasonCA @ Jan 6 2013, 16:16) *
So whether the audio source is vinyl records or audio cossets or otherwise, does capturing audio data at a slower analog playback speed effect the final audio quality? And, if so why? Or, does it result in essentially the same? Or, is the audio quality even better?


Unless I missed it, I didn't see anyone comment on the common practice of using half-speed to improve high frequency response. When a record is played at a slower speed, the cartridge can more easily track high frequencies with less "spitting" type distortion. But yes, RIAA (records) and NAB (tape) curves will not be accurate.


The trade off is that lower rotational speeds may hurt dynamic range by reducing cartridge output.

IME a good modern cartridge (e.g. Shure M97+ alphabet soup) in a good tone arm and TT (in my case an older Rega) will avoid audible problems with high frequency distortion and frequency response losses.

QUOTE
Personally, if I had old records or tapes that I recorded digitally, I'd probably EQ them by ear to sound as I like. This could possibly improve on the original tonality, and would also compensate for improper RIAA and NAB playback response curves.


I'd definitely experiment with that if I didn't like the sound of something after doing a straight-forward job of transcribing it. In many cases that works.
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 22nd October 2014 - 04:51