If you had a good SRC would you go in upsampling

[MOB-1207]

Dear all ME,

If you had a good SRC, would you do your Mastering in Upsampling without a doubt?

I have Saracon in the Mastering studio where I am working. There's a lot of documents about the Upsampling but would you take the time to upsample all your files in high resolution and after the AD make the downsample? Anyone doing this?

Thanks a lot

[PureArtist]

I would go downsampling without a doubt, up-sampling with color maybe, you ask how ? send the sound to a high-end av amplifier then record the output of it to the higher sample rate you want. Oh & the color wil be super color full, lol.

[MOB-1207]

Originally Posted by PureArtist

I would go downsampling without a doubt, up-sampling with color maybe, you ask how ? send the sound to a high-end av amplifier then record the output of it to the higher sample rate you want. Oh & the color wil be super color full, lol.

I Never saw an answer like this! You're a king of genious and I want to thank you so much for the answer... So... Are you STILL working in a grocery?, lol

[dcollins]

Originally Posted by MOB-1207

Dear all ME,

If you had a good SRC, would you do your Mastering in Upsampling without a doubt?

I have Saracon in the Mastering studio where I am working. There's a lot of documents about the Upsampling but would you take the time to upsample all your files in high resolution and after the AD make the downsample? Anyone doing this?

There are arguments either way. You have to try it yourself and see if you prefer the upsampled/downsampled version. If all you are doing is EQ, there will be no positive difference. With digital compression or limiting, there is a potential advantage to it.

One thing to remember is that the process of upsampling only adds noise and distortion to the music, it does not increase the "resolution" of the signal.


DC

[MOB-1207]

Originally Posted by dcollins

There are arguments either way. You have to try it yourself and see if you prefer the upsampled/downsampled version. If all you are doing is EQ, there will be no positive difference. With digital compression or limiting, there is a potential advantage to it.

One thing to remember is that process of upsampling only adds noise and distortion to the music, it does not increase the "resolution" of the signal.


DC

Thank you very much for your advice DC. Also, thank you for the amazing job that you did on the soundtrack of "The village" by James Newton Howard. I listen to it and the sound is a reference for me. Dynamic is perfect and depth too.

[aivoryuk]

Originally Posted by dcollins

If all you are doing is EQ, there will be no positive difference. With digital compression or limiting, there is a potential advantage to it.



DC

This is why I like Wavelab and the crystal resampler as I often insert it before the compressor as it can sound smoother but it really depends on what I am working on.

FWIW I don't use the crystal resampler for downsampling as I thinki there are better ones out there, but for upsampling I think it is more than fine.

[jayfrigo]

Originally Posted by MOB-1207

Dear all ME,

If you had a good SRC, would you do your Mastering in Upsampling without a doubt?

I have Saracon in the Mastering studio where I am working. There's a lot of documents about the Upsampling but would you take the time to upsample all your files in high resolution and after the AD make the downsample? Anyone doing this?

Thanks a lot

Could you explain a bit more about where you would upsample in the signal chain? If you're using an A to D, why not simply sample at the higher rate? Do you mean upsample before the DAC feeding the analog chain? That wouldn't be of much use. And upsampling after A to D makes no sense when you could just sample at the higher rate to begin with. If it's an all digital chain, I'm not sure why you mentioned the "AD," but in that case, as DC says, non-linear processes may be helped by working at higher rates.

[CharlesCola]

Not knowing digital stuff too deeply I read what seems to be a credible paper on upsampling before DA conversion and upsampling DA converters in general: http://www.mlssa.com/pdf/Upsampling-theory-rev-2.pdf

I read this after extensive AB tests with a Lucid 88192. I was testing the converter when doing DA at the project sample rate of 44.1khz and then turning on the Lucids SRC so that it was running at 192khz. It was obviously much better sounding at 192khz on all material sent to it.

Following this I researched alot to find out why this could be and found the paper above which describes how the converter will send out subjectively better audio when running at 192hz because of reasons I don't even begin to understand (aliasing from filtering will occur outside audible bandwidth etc... which seems the opposite of what dcollins says) but seem logical.

The Lucid uses a hardware SRC chip on every AES input - the AD1895. This is a common device used in many high end "Audiophile" upsampling dacs.

Either way IMO the Lucid definitely sounds much better when upsampling....

[dcollins]

Originally Posted by CharlesCola

Not knowing digital stuff too deeply I read what seems to be a credible paper on upsampling before DA conversion and upsampling DA converters in general: http://www.mlssa.com/pdf/Upsampling-theory-rev-2.pdf

I only skimmed it, but looks like OSAF.


DC

[Bob Olhsson]

My experience has been sometimes it's better and sometimes it's worse. My guess is that this depends a lot on the converters used for the recording. My best results have been with the recordings that used the cleanest converters.

I realize that's not much help but we're talking about how one extra process interacts with a whole system of signal processing where less can often be more.

[jayfrigo]

Originally Posted by CharlesCola

The Lucid uses a hardware SRC chip on every AES input - the AD1895.

In the pro world, the SRC chips in these types of DAC designs are often used for jitter attenuation rather than upsampling.

Remember, upsampling can not create data that was never sampled in the first place (meaning nothing above the original Nyquist frequency), nor does it make any "stair steps" smaller since the anti-image filter precludes their existence at the output regardless.

[CharlesCola]

Is there any truth in the ideas/opinions stated it the paper above (http://www.mlssa.com/pdf/Upsampling-theory-rev-2.pdf) that expresses the subjective sonic improvement is due to more than just jitter reduction (i.e. as well as less jitter could there also be less aliasing/artifacts in audible spectrum due digital processes/filtering etc taking place at higher sample rate)?

[dcollins]

Originally Posted by CharlesCola

Is there any truth in the ideas/opinions stated it the paper above (http://www.mlssa.com/pdf/Upsampling-theory-rev-2.pdf) that expresses the subjective sonic improvement is due to more than just jitter reduction (i.e. as well as less jitter could there also be less aliasing/artifacts in audible spectrum due digital processes/filtering etc taking place at higher sample rate)?

J*tter is only an issue at the points of analog conversion. It has no effect on DSP whatsoever.

The claims of less "time smear" in up-sampled signals being the audible factor is by no means settled, imo.

It is correct that asynchronous SRC can be used for j*tter attenuation before D/A conversion.


DC

[Bob Olhsson]

Originally Posted by dcollins

It is correct that asynchronous SRC can be used for j*tater attenuation before D/A conversion...

My understanding is that this is mostly a matter of the very best SRC chips having better jitter reduction and filters than most DAC chips. The down side is significantly greater latency.

[dcollins]

Originally Posted by Bob Olhsson

My understanding is that this is mostly a matter of the very best SRC chips having better jitter reduction and filters than most DAC chips. The down side is significantly greater latency.

ASRC is a catch-22, as it acts as a total firewall for j*tter, but at the expense of changing the data (although the error is far below audibility) and as you said the latency is always higher. I have experimented with just a good quality PLL and the SRC and I think I'm going to put a switch to bypass the SRC as in some applications is might not be needed.

DC