anyone use a upsampler?

K

kool bubba ice

Limp Gawd
Joined
Sep 18, 2008
Messages
328
It adds bits that are not there, if your DAC supports 24 bit, and upsamples to 192 hz. Someone whose opinion I trust, told me its like upconverting players, adding pixels that aren't there to offer a smooth, rich, detailed picture. I'm interested in getting one. Plus, the Jitter control is very good.
 
There is no interesting benefit to upsamping. A greater sample rate may reduce jitter to a certain extent, but only if the converters don't perform oversampling internally (most DACs produced during and after the 90's do), and only if the DACs don't clock off another source entirely (such as an external master crystal clock).

Another factor that skews the potential quality of an upsampled signal is non-integral interpolation. When you upsample a 44.1 kHz signal to 192 kHz, you're attempting to change one sample into 4.3537414965986394557823129251701 samples, and that's not a very pleasing thing to try and do to an audio stream that was perfect to begin with. The level of imprecision involved in the interpolation will yield potentially measurable quantization distortion. Oversampling DACs perform integral interpolation, which is much "nicer" from a mathematical perspective, but they still must rely upon additional filters to avoid potential artifacts in the resultant signal. There's a reason mastering engineers tend to request mixes delivered at 88.2 or 176.4 kHz.

As for pushing a 16-bit word into a 24-bit word, there's zero benefit. All you're doing is adding eight bits of padding per word. You could go from 16-bit to 896-bit and get the exact same benefit: absolutely nothing.

To add a cherry to the top of the disappointing upsampling sundae, the vast majority of listeners can't tell any difference between an actual 24-bit, 192 kHz signal, with actual musical content up to ~96 kHz, and the 16-bit, 44.1 kHz CD counterpart. There's nothing audibly "smoother, richer, more detailed" about the original higher-res signal.

The first rule about digital audio: don't fuck with it. Don't upsample it, downsample it, convert the bit depth or do anything else at all with it. Just spit it out, bit-perfect if you really, really want to, but don't even sweat that. Your DAC may oversample it, but that's only there to make it less expensive to make, not to increase the quality in any meaningful way.

 
Very thorough. So any audible difference is Placebo? Why do upconverted pictuers look better? They are adding pixels that are not there. I know it's video and not audio, but still seems relevant to me.
 
Phide: Out of curiousity...what is your setup? Do you use a DAC or just use analog or digital out to your receiver/amp?
 
This article explains sampling rates and bit depth in easy to understand language. I've never noticed any smoother sound quality when I tried 96khz 24bit.

http://books.google.ca/books?id=4-2...&hl=en&sa=X&oi=book_result&resnum=6&ct=result

Read this to actually find out what is the best sampling rate to use with your specific hardware.

http://www.hydrogenaudio.org/forums/index.php?showtopic=9772&hl=udial

I found 44.1khz and and 48khz gave me the same results but I use 44.1khz 16bit with dithering in foobar anyway.

Using UDIAL I found that I got some distortion when using 88.2khz and 96khz but none with 44.1khz or 48khz. Read the warnings about using it so you don't damage your speakers. I used a pair of cheap IEMs to run this test.
 
kool bubba ice said:
Very thorough. So any audible difference is Placebo? Why do upconverted pictuers look better? They are adding pixels that are not there. I know it's video and not audio, but still seems relevant to me.
Click to expand...
With video, you're not at the point of diminishing gains when you upconvert a DVD or SD broadcast to HD. The differences are sometimes subtle, but a good interpolating upconverter can actually mangle the signal a fair bit to achieve a result that looks good, and that isn't really feasible with an audio stream. With audio, particularly 16-bit, 44.1 kHz CD audio, you're already at the point where any improvement in unlikely to be audible.

The process of upsampling may be a little bit more critical to 11.025 kHz or 22.05 kHz signals, but the limiting factor with those sampling rates is the Nyquist limit, not aliasing. Interestingly enough, if you took an original 44.1 kHz file, lowpassed it to 10.5 kHz or so and downsampled it to 22.05 kHz, odds are you wouldn't be able to tell the difference between the original 44.1 lowpassed file and the 22.05 file. The sample rate for CD audio is so "high" only because it gives us a good ~22 kHz Nyquist limit, so we end up with the full audible spectrum.

450 said:
Phide: Out of curiousity...what is your setup? Do you use a DAC or just use analog or digital out to your receiver/amp?
Click to expand...
Straight out of my Prelude's analog outs to a Headroom Micro to the HD650. Nothing too fancy.
 
Pretty much everything posted above is wrong. Go listen to a real upsampling transport or processor like a dCS Purcell and report back.
 
Oh? Show me some double-blind A/B testing for that Purcell, showing any observed differences between it and any other non-upsampling DAC on the market (not those that oversample), and report back. Unless, of course, you believe that non-blind listening trials are a valid way to prove basic science wrong.
 
I use an upsampling plug-in with Foobar. However, that's just so that I don't have to worry about sampling rate with ASIO output.
 
phide said:
Oh? Show me some double-blind A/B testing for that Purcell, showing any observed differences between it and any other non-upsampling DAC on the market (not those that oversample), and report back. Unless, of course, you believe that non-blind listening trials are a valid way to prove basic science wrong.
Click to expand...

I've done my own A/B blind tests it is not subtle. I do not believe in upsampling for the sake of upsampling, but when properly implemented an upsampler can interpolate the data from a CD in such a way that the end result sounds more like the original master tape.

I upsample to 44.1 CD to DSD and it is like listening to a different format.

I would not argue that a player or DAC that claims to upsample is always going to be better than a nonupsampling setup, however.
 
Blackstone said:
I've done my own A/B blind tests it is not subtle. I do not believe in upsampling for the sake of upsampling, but when properly implemented an upsampler can interpolate the data from a CD in such a way that the end result sounds more like the original master tape.
Click to expand...
If what you claim is accurate, and if we assume for the sake of the basis of the argument that the particular upsampler in question achieves an upsampled result that's nearly identical to the original 192 kHz master (if such a thing existed -- seldom the case in music production), then obviously you'd have no trouble discerning an original 192 kHz sample from a sample downsampled to 44.1 kHz using a linear phase resampler, correct? Such a test would isolate your specific claim here that proper interpolation provides such stunning results, so any picosecond variations in clocking jitter from upsampler to upsampler and from DAC to DAC wouldn't be factored in.

If you're as passionate about the benefits of upsampling as you seem to be, certainly you'd have no objection to putting your money where your mouth is. You'd have no issue with such a test given that the downsampled file would have distortion characteristics not present in the original, making it all the easier for you, right?

I can provide original 192 kHz samples if you need them, and I'm also willing to resample them using the best resampler I have available if you don't have the means.
 
It's gonna run at whatever the bitwidth of your DAC is anyway. Depending on how well your DAC handles lower (or higher) bitrate input, it will do one of:nothing, give a worse result, give a better result. Note that the cases where there is a different result will probably be very small in most cases.

Software upsampling would allow you to use an arbitrary algorithm with an arbitrary degree of complexity and precision. How necessary that is? Well...usually not very. And it will still be limited by the original source resolution...in ALL cases you would do better by using a higher source bitrate to begin with if possible.
 
I certainly notice the difference between 192khz and 44.1khz on my $10k speaker system:D...
 
Yea, I was wondering what the point is of using software to increase bit depth to 24bit would be when my EMU 0404 USB manual says it it internally converts to 32bit in hardware anyway.
 
phide said:
If what you claim is accurate, and if we assume for the sake of the basis of the argument that the particular upsampler in question achieves an upsampled result that's nearly identical to the original 192 kHz master (if such a thing existed -- seldom the case in music production), then obviously you'd have no trouble discerning an original 192 kHz sample from a sample downsampled to 44.1 kHz using a linear phase resampler, correct? Such a test would isolate your specific claim here that proper interpolation provides such stunning results, so any picosecond variations in clocking jitter from upsampler to upsampler and from DAC to DAC wouldn't be factored in.

If you're as passionate about the benefits of upsampling as you seem to be, certainly you'd have no objection to putting your money where your mouth is. You'd have no issue with such a test given that the downsampled file would have distortion characteristics not present in the original, making it all the easier for you, right?

I can provide original 192 kHz samples if you need them, and I'm also willing to resample them using the best resampler I have available if you don't have the means.
Click to expand...

I like the idea of your experiment. I believe the Purcell does have downsampling capabilities--for example it can take DSD (SACD) and downsample to 44.1 for use with a conventional 44.1 DAC.

I can't do that experiment right now because my transport (Squeezebox) only transmits 44.1. I have no way of playing the original 192kHz sample or real DSD at the moment. I might be able to rig something in the future when I move to a USB based transport. Frankly, however, I'm really not interested. My claims are not about upsampling in general, but rather about the Purcell, which is an upsampler.
 
You must log in or register to reply here.
Back
Top