|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
66.68.73.9
I assume that since my DAC does not oversample, then SRC won't work. Am I correct?
Follow Ups:
This is a complex subject, I did a lot of experimentation on it last year, here is what I found:Upsampling to a DAC with a digital filter (ie an oversampling DAC) usually makes them sound better. The primary reason is that the brickwall filter frequency is directly related to the sample frequency, if the sample frequency goes up, so does the filter frequency. Getting the filter higher up above the audio band almost always helps.
You can change this sound significantly by adjusting the upsampling frequency and how its done. An integer upsampling almost always sounds better, so use 88.2 rather than 96 for CD files. (this is all assuming the upsampling is NOT being done by an ASRC chip, those have their own problems). Even at 88.2 you can get changes in sound by choosing different upsampling algorithms. I'm not going to tell you which is the best, its a very personal thing. I've done the tests where two people in the same room, same system etc. prefered different algorithms.
I've yet to run across a situation where going to higher than 88.2 for CD improves the sound. The only reason I would ever use 192 is if I had a true 96 original, and even then I'm doubtful about the outcome. Yes the modern chips are all supposed to run at 192, but every measurement I've made with chips running at 192 has been worse than with 88.2/96. The jitter goes up, distortion goes up etc. As far as I can tell this is because there are a number of timing parameters in the chips that are fixed, as the sample rate goes up these wind up being larger percentages of the sample time.
For a DAC without a digital filter (ie a NOS DAC) it gets much more subtle. Using software upsampling CAN improve the sound but its going to be very DAC and system dependant. It can also degrade the sound!
The common NOS DAC has no filtering either digital or analog after the DAC chip, relying on the downstream components or the human ear to do the filtering. This may or may not be an issue. Upsampling will move the ultrasonic frequencies up higher which MIGHT make an audible improvement. If the DAC chip being used only has 16 bit resolution (such as the infamous 1543 DACs) The upsampling cannot "smooth out" the waveform so it might actually make the sound worse. If your NOS DAC uses a bit more modern chip that can handle greater than 16 bits and is designed for higher frequencies upsampling could actually improve the sound.
Again I would not use 96 for CD sound but stick to 88.2.
With NOS DACs I've been finding out that one of the things that makes them sound the way they do is that a lot of DAC chips sound better when run at slow speeds. Its those fixed timing parameters again. If you don't have any oversampling then those fixed parameters are a very tiny fraction of the sample period, much less than with the traditional 8X oversampling.
So a 2X upsampling (88.2) can push the ultrasonics up a lot, which can significantly improve the sound for some systems/people, and is not decreasing the sample period THAT much so the fixed timing parameters are not degrading the sound significantly.
Its all a balancing act!
Your DAC needs to support higher bit rates like 24/96 etc. SRC and other upsamplers will change the file to a higher bit rate which is then passes to a dac.If your dac does not support these rates you will get nasty white noise or silence.
No, it depends on the DAC chip used in your DAC. Most newer DAC chips can do 24/96 or even 24/192, including the Philips 153X series, AD1853, and CS4396/7.
Yes, but, just because a DAC an support higher frequencies doesn't mean it is going to sound any better (and could certainly sound worse) with a higher input frequency.It is important to stress that using resampling does nothing whatsoever to improve an audio signal. It will often result in a degradation of the original data, which is why some people prefer the SRC resampler, because it apparently damages the data the least.
Upsampling only makes sense to me if you are using a DAC that actually sounds better with higher frequency inputs. The Benchmark DAC1 is an excellent example of such a DAC because it automatically resamples all of its inputs to 110K (and doesn't do as good a job as the SRC resampler, so using SRC to at least get the data to 96K so the DAC1 can mangle it a bit less when converting to 110K makes sense). The DAC1 is the only DAC I've used that actually sounds better with resampling.
It should be noted that one of the reasons upsampling is likely to degrade your audio performance (with the exception of the DAC1, for the reasons described above) is that it puts further strain on the I/O systems of your transport and DAC (not to mention your computer's CPU, if you are using a computer to do the upsampling, which could have an undesirable effect on the computer's ability to transmit the data at the appropriate speed).
Scrith-"The Benchmark DAC1 is an excellent example of such a DAC because it automatically resamples all of its inputs to 110K (and doesn't do as good a job as the SRC resampler, so using SRC to at least get the data to 96K so the DAC1 can mangle it a bit less when converting to 110K makes sense)."Does the Benchmark DAC1 force the user to upsample to 110K? There are no 88.2 or 96kHz settings?
Just wondering. The thought of resampling from 44.1 to 96 to 110... well it just *seems* like a bad idea... no? lol
Anyways, I wonder... with THAT DAC if skipping SRC and just going 44.1 straight in is the way to go.
Ten years ago 90% of audiophiles might have said "24/96 is more better" but now people (you included) are saying - sure we get the nyquist frequency up which allows the brickwall filter to be higher as well - but what are the sonic penalites?
Very good approach.
So when is Benchmark coming out with a multichannel USB DAC? :D
The Benchmark DAC1 resamples all inputs to 110K internally. The user is free to use whatever sample rate they want (44.1 most of the time, I assume, but it supports inputs of up to 192).Yes, it does seem like a bad idea on the surface. But this is how they reduce the effect of jitter. I've owned and used Benchmark DAC1s for a couple of years and I can assure you they sound fantastic. I am currently using a Lavry DA10, which sounds very similar, perhaps a bit better to my ears, but it any event it allows me to use a jitter-prone source (computer) without resampling (which I think is a bad thing, generally).
If Benchmark ever came out with a DAC with USB input (which didn't convert to S/PDIF internally, as it should be done) I would probably buy it. If it didn't need to resample to 110K I would really want to purchase it. And if they came up with their own USB driver that supported some kind of asynchronous I/O (or a similar driver that used an Ethernet or Firewire interface) I would be the first one in line with the cash for it!
Scrith:I want a multichannel USB solution that is an 8-channel DAC c/w volume control that a)does not use USB Bus power and b)does not convert to S/PDIF but instead use I2S internally.
I know Gordon Rankin could solder me up a one-of-a-kind unit... but we all know how much the cost of a product is when you're only making ONE! :D
Ah, dare to dream!
The other way to go is to "modify" the Foobar output and crossover plugins so that you can map to different audio devices simultaneously, thus negating the need for a multichannel product - you could then just use multiple stereo pairs. The trouble is... if you have three identical audio devices they will show up with the same device name, unless you create a custom installation inf file for each separate intall, thus giving you the power to have unique names.
A good USB solution can be had - but if you're going the ACTIVE route like I always seem to end up doing... that's a whole other shooting match!
... because the stop band is increased from 20K all the way up to the new sample rate, instead of being in the narrow band from 20K to the CD sample rate Nyquist frequency (22.05K). So apparently you get can get rid of brickwall filtering just by upsampling, which on the surface seems too good to be true :)Of course you are right that nothing new is created, and there will always be some degradation, but getting rid of the super steep filter could make a difference.
...I think I remember hearing that increasing the sample rate can move jitter artifacts out of the audible hearing range. Or something like that. I have a bunch of sound cards and usb interfaces and I can't tell the difference with most of them if I resample with src.
This post is made possible by the generous support of people like you and our sponsors: