Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Matt_G]

Ok we've moved this discussion from Brad Blackwood's forum to here with the hope of getting some definitive answers from the experts here.

The debate is split into 2 areas which are this...

1) When mastering for CD format & the client brings in 24bit 44.1kHz mix files is there any sonic benefit to had by upsampling & processing at 88.2 or 96kHz, even though the final format will be downsampled to 16bit 44.1kHz at the end? The main concern here is whether the sample rate conversion artifacts will undo or outweigh any of the sonic benefits from working at higher sample rates in the first place.

2) When mastering completely in the digital domain is there any benefit at using higher sample rates at all when the final destination is CD format (16bit 44.1kHz)?

One other thing I've often wondered & is worth mentioning in the context of this thread, is when working in the analog domain & then capturing to digital at the end. Is it better to capture at 24bit 96kHz or 16bit 44.1kHz if there is no further processing to be done, therefore avoiding an unnecessary SRC process?

Obviously the quality of the sample rate conversion is extremely important in the context of this discussion, so presume we are talking about using very high quality SRC like the Weiss SRC-1 or R8Brain Pro for example.

Matt

[PookyNMR]

If you are using plug-ins to do digital processing, some manufacturers state that their products perform better at the 2x sample rates.

[Matt_G]

PookyNMR wrote on Thu, 29 June 2006 06:36

If you are using plug-ins to do digital processing, some manufacturers state that their products perform better at the 2x sample rates.

Yes the Massenburg Design Works MDWEQ upsamples 2 x if it's used in a 44.1kHz or 48kHz session. But do you think this benefit will translate when downsampled to 44.1kHz at the end of the mastering process? If so why?

Matt

[cerberus]

yes i've heard increased fidelity to the souce from upsampling; i've observed it many times. i don't know where the point of dimishing returns lies; but i think my ears are not lying to me.  i posted that observation and a specific example (which i invite anyone to try and repeat) in the related thread on this forum.

jeff dinces

[danlavry]

cerberus wrote on Thu, 29 June 2006 17:52

yes i've heard increased fidelity to the souce from upsampling; i've observed it many times. i don't know where the point of dimishing returns lies; but i think my ears are not lying to me.  i posted that observation and a specific example (which i invite anyone to try and repeat) in the related thread on this forum. jeff dinces

I do not have a problem with someone upsampling, doing some processing then down sampling back. There many way to do things, and at times going X2 or even way up is called for, or provides an advantage.

It is difficult to have a good DA anti imaging analog filter, so we upsample way up.

It is often easier to process near 20KHz with 88.2-96KHz, so going X2 may make sense.

And so on...

But that has nothing to do with the basic sample rate, which determines what music you capture. That should relates to the range of frequencies one can hear. Upsampling does not alter what you can hear, it may make some specific localized tasks work better.

I said many times before. I am sorry it is not simple or easy to understand the distinction between audio sample rate (the rate for conversion, transfer, storage) and "localized rates" for specific reasons. Hack, I know of cases that would require localized upsampling to near 10MHz rates... Followed, of course, by serious down sampling.

Regards
Dan Lavry
www.lavryengineering.com  

[cerberus]

danlavry wrote on Thu, 29 June 2006 13:54

It is often easier to process near 20KHz with 88.2-96KHz, so going X2 may make sense.

that was my initial reason, but i find that even for the simplest dsp processes (e.g. summing) that there is a noticible tightening up of bass as well. perhaps bass instruments make enough hf content that a filter acting near 20khz touches them? not sure why that is the case on my system, but it sealed my decision to upsample 44.1/48khz sources as a rule before attempting any dsp whatsoever.

note that i chose my [software]src carefully, something you alluded to on the other thread but i think needs emphasis.  if i were using an src that apple or steinberg or digidesign provided me with, then i would instead be observing alarmingly clear sounding...artifacts   .

jeff dinces

[Patrik T]

Say we have a 44.1 source and want to process it digitally and want to use three plugins, which all have an option to upsample 2x for 44.1/48 material.

If we SRC to, say, 88.2 before this operation starts, there will be one upsampling. And later on a downsampling back to 44.1. But, we route 88.2 through the plugs and choose to not let them upsamle internally.

If we do not upsample the source and choose to run the plugs in upsampling mode, won't there be SIX (up & back in each plug)instances of SRC, rather than TWO (up before plugs and back after plugs)?

If a analogue source is recorded at 96 kHz and there are plans to use digital tools on the digital audio image of the analogue source, there would be only one downsampling, if we look at the scenarios above.

Maybe Jon Hodgson will chime in with his good thoughts on this.

Best Regards
Patrik

[Jon Hodgson]

Patrik T wrote on Fri, 30 June 2006 14:12

Say we have a 44.1 source and want to process it digitally and want to use three plugins, which all have an option to upsample 2x for 44.1/48 material. If we SRC to, say, 88.2 before this operation starts, there will be one upsampling. And later on a downsampling back to 44.1. But, we route 88.2 through the plugs and choose to not let them upsamle internally. If we do not upsample the source and choose to run the plugs in upsampling mode, won't there be SIX (up & back in each plug)instances of SRC, rather than TWO (up before plugs and back after plugs)? If a analogue source is recorded at 96 kHz and there are plans to use digital tools on the digital audio image of the analogue source, there would be only one downsampling, if we look at the scenarios above. Maybe Jon Hodgson will chime in with his good thoughts on this. Best Regards Patrik

You're basically right, reducing the number of critical SRC processes is generally a good thing, plus there is a high probability that a stand alone or mastering SRC is going to be higher quality than one built into a plugin due to the differing priorities - the developer can afford to throw more cpu resources at it.

However it's not neccessarily an either/or mechanism, which is why I referred to minimizing critical SRC processes, which I am using to refer to the last stage, that takes you to your final output rate.

If a non linear plugin gains from being run internally at 96kHz on a 48kHz stream, due to any aliasing artifacts being reduced by the extra frequency headroom, it may benefit further from being run at 192kHz internally (assuming of course there is provision for this)

The lesser quality of the internal SRC of the plugin would be less of an issue in this case, since if we assume that most artifacts (aliasing and frequency response imperfections) will be up near nyquist, this is the frequency area which will be eliminated in your final mastering SRC (and anyway you can't hear it even if you don't get rid of it).

[Patrik T]

Thank you Jon!

[Pingu]

cerberus wrote on Fri, 30 June 2006 05:42

danlavry wrote on Thu, 29 June 2006 13:54 It is often easier to process near 20KHz with 88.2-96KHz, so going X2 may make sense. that was my initial reason, but i find that even for the simplest dsp processes (e.g. summing) that there is a noticible tightening up of bass as well. perhaps bass instruments make enough hf content that a filter acting near 20khz touches them?

Im noticing this too.

[Matt_G]

danlavry wrote on Fri, 30 June 2006 03:54

I do not have a problem with someone upsampling, doing some processing then down sampling back. There many way to do things, and at times going X2 or even way up is called for, or provides an advantage. It is difficult to have a good DA anti imaging analog filter, so we upsample way up. It is often easier to process near 20KHz with 88.2-96KHz, so going X2 may make sense. And so on... But that has nothing to do with the basic sample rate, which determines what music you capture. That should relates to the range of frequencies one can hear. Upsampling does not alter what you can hear, it may make some specific localized tasks work better. I said many times before. I am sorry it is not simple or easy to understand the distinction between audio sample rate (the rate for conversion, transfer, storage) and "localized rates" for specific reasons. Hack, I know of cases that would require localized upsampling to near 10MHz rates... Followed, of course, by serious down sampling.

Thanks Dan, it is important to note the differences between recording at high sample rates & upsampling to higher sample rates to clean up non linear processes. Thanks for clearing that up.

So what about capturing the final output from an analog mastering path at 16bit 44.1kHz, compared to capturing at 24bit 96kHz when the final medium is going to be CD? In this case is it better to capture at a high resolution then downsample & dither to 16bit 44.1kHz or simply capture at 16bit 44.1kHz to avoid a further DSP step that an SRC & dither would require? Or perhaps 24bit 44.1kHz would be a good compromise? Any thoughts on this Dan?

Matt

[danlavry]

Matt_G wrote on Thu, 06 July 2006 13:55

danlavry wrote on Fri, 30 June 2006 03:54 I do not have a problem with someone upsampling, doing some processing then down sampling back. There many way to do things, and at times going X2 or even way up is called for, or provides an advantage. It is difficult to have a good DA anti imaging analog filter, so we upsample way up. It is often easier to process near 20KHz with 88.2-96KHz, so going X2 may make sense. And so on... But that has nothing to do with the basic sample rate, which determines what music you capture. That should relates to the range of frequencies one can hear. Upsampling does not alter what you can hear, it may make some specific localized tasks work better. I said many times before. I am sorry it is not simple or easy to understand the distinction between audio sample rate (the rate for conversion, transfer, storage) and "localized rates" for specific reasons. Hack, I know of cases that would require localized upsampling to near 10MHz rates... Followed, of course, by serious down sampling. Thanks Dan, it is important to note the differences between recording at high sample rates & upsampling to higher sample rates to clean up non linear processes. Thanks for clearing that up. So what about capturing the final output from an analog mastering path at 16bit 44.1kHz, compared to capturing at 24bit 96kHz when the final medium is going to be CD? In this case is it better to capture at a high resolution then downsample & dither to 16bit 44.1kHz or simply capture at 16bit 44.1kHz to avoid a further DSP step that an SRC & dither would require? Or perhaps 24bit 44.1kHz would be a good compromise? Any thoughts on this Dan? Matt

Hi Matt

I would certainly separate the number of bits from the sample rate.

First about bits:
In the good old days, we were struggling to have clean 16 bits. The analog noise floor was so high, that it served as dither. At very high noise, referenced to the quantization level, the dither action was fine, but the noise penalty (reduced dynamics)was too high.

Next we started having the noise approaching the 16 bit quantization, eventually having the noise lower then the 16 bit quantization. At that point, it became necessary to do something about the unwanted "art effects" (distortions and noise modulation) of low level signals. By now, many AD's are better then 16 bits. In fact, when an AD is a "true 18 bits" or more, and the end result is say 16 bits, it is best to dither. There is no reason to record directly to 16 bits, when you can record to 24 bits (the last few bits are of no value, they are noisy) and then dither to 16 bits.

Of course, a noise shaped dither is better then the older varieties, because the added noise resides in frequencies where the ear sensitivity is lower.

As a rule, keep the word length as wide as you can (say 24 bits), and dither at the last step of the processes (with a noise shaping type if you can).

Regarding the sample rate: take 44.1KHz vs. 88.2KHz. This days, virtually all converters "do their job internally" at much higher rates then 44.1 or 88.2... Therefore, having 44.1KHz requires some down-sampling to take place. So the first question that come to mind is: which is a better downsampler? The one in the AD? An external SRC? This is important because the last stage in the down sampling process is the one performing 88.2 to 44.1KHz. That last stage is the most difficult and demanding one (in terms of down sampling). So the answer depends on specific implementations of the SRC...

Regards
Dan Lavry  
http://lavryengineering.com  

[blue2blue]

Dan, It's possible I've misread your comments elsewhere, but I've got the impression you have suggested that if one wants to record and work at a 'higher' rate but move down to a final product at 44.1 kHz (say for a CD) via SRC  -- that it's better to pick a "sychronous rate" -- that is,  one that's an even multiple of the target rate.

(In this case, 88.2 kHz as opposed to 96 kHz. I know 88.2 kHz is closer to your suggested 'optimal sampling rate' in the 60-70 kHz   range -- but in this case I'm focusing on the potential 'degradation' resulting from SRC from an asynchronous rate.)

Is that the case or have I grossly oversimplified or neglected or misunderstood something?

Thanks in advance.

[Pingu]

Youve got the jist of it but i still dont buy into the even integer
samplerate debate yet.

It sounds logical but processing audio does not always follow this path.

[Jon Hodgson]

Pingu wrote on Sun, 09 July 2006 16:50

Youve got the jist of it but i still dont buy into the even integer samplerate debate yet. It sounds logical but processing audio does not always follow this path.

I can't see that there is any debate, if we're talking about signal accuracy as opposed to possible subjective preference.

If your constraint when designing your filter is memory for coefficients, then the 2:1 samplerate conversion will be superior, because it can use half the available coefficients per phase. Any other ratio requires more phases, and therefore fewer coefficients per phase.

If your constraint is processing cycles, then the worst case scenario is that the 2:1 conversion is equally good.

The truth of the matter however is that given a good quality converter and high quality SRC I think it's unlikely that people could pick out the difference, especially when dealing with real music.