Is it worth outputting @192 KHz 24 bit from Media PC for music?

Discussion in 'Hi-Fi Stereo Systems & Separates' started by Dean, Feb 9, 2009.

  1. Dean

    Dean
    Well-known Member

    Joined:
    Jun 20, 2003
    Messages:
    3,034
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    106
    Location:
    London
    Ratings:
    +347
    Hi, I have recently been archiving my CD collection onto my Sony Vaio VGX-XL301 via WMA lossless format, the PC can output at the following sampling rates to my STR-DA3200ES:

    2 Channel, 16 bit, 44.1 KHz (CD quality)
    2 Channel, 16 bit, 48 KHz (DVD quality)
    2 Channel, 16 bit, 96 KHz (Studio qulaity)
    2 Channel, 16 bit, 192 KHz (Studio quality)
    2 Channel, 24 bit, 44.1 KHz (Studio quality)
    2 Channel, 24 bit, 48 KHz (Studio quality)
    2 Channel, 24 bit, 96 KHz (Studio quality)
    2 Channel, 24 bit, 192 KHz (Studio quality)

    PS Using Windows Vista Home Premium (SP1) and Windows Media Center as a front end.

    I usually stick to 2 channel, 24 bit, 48 KHz for my STR-DA3200ES to be able to decode 2ch into multichannel (PLII), but is it really worth switching to 2 channel, 24 bit, 192 KHz when just listening to music, in theory I guess it should but then I heard it depends how the CD was sampled/mastered?

    Cheers for any feedback!
     
    Last edited: Feb 10, 2009
  2. Trollslayer

    Trollslayer
    Distinguished Member

    Joined:
    Feb 11, 2007
    Messages:
    26,079
    Products Owned:
    0
    Products Wanted:
    1
    Trophy Points:
    166
    Location:
    Poole
    Ratings:
    +10,932
    I don't see the point of oversamping for recording purposes, 44.1kHz 16 is what's on the CD.
     
  3. Dean

    Dean
    Well-known Member

    Joined:
    Jun 20, 2003
    Messages:
    3,034
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    106
    Location:
    London
    Ratings:
    +347
    Thanks for the reply, I should have said that the sampling outputs quoted relate to playback, the CDs have been recorded/archived via wma lossless format (mathematically lossless). So for playback is it worth oversampling?
     
  4. mr_yogi

    mr_yogi
    Active Member

    Joined:
    Sep 12, 2003
    Messages:
    1,640
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    66
    Ratings:
    +173
    From what I've read about PC (Windows) audio, I'd be more concerned with taking the Windows "k mixer" (or whatever it's called) out of the chain.
     
  5. deaf cat

    deaf cat
    Active Member

    Joined:
    Feb 9, 2005
    Messages:
    1,224
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    48
    Ratings:
    +58
    If you hear a difference and you like it then yes....

    Maybe download a track or two that are 24bit 96KHz have a listen, down sample them to 16bit 44.1KHz and listen again.
     
  6. Steven

    Steven
    Senior Moderator

    Joined:
    Feb 1, 2005
    Messages:
    37,620
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    166
    Ratings:
    +6,604
    :) Depends on each individual

    Some people claim to hear a "distinct difference". Some people claim to "hear a difference when listening intensely"

    Me with my Edirol UA-25 sound card and foobar... hmm erm... what now? :laugh:

    Stopped caring long ago. Just have it on standard 44.1 and went back to enjoying the music
     
  7. alexs2

    alexs2
    Well-known Member

    Joined:
    Nov 13, 2002
    Messages:
    13,967
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Ratings:
    +1,671
    Much better idea....as far as I know,Windows samples everything to 48kHz and generally does that badly.

    To then resample to 96kHz or above introduces yet another layer of processing,and certainly removing KMixer from the chain by using something to bypass it would be a much better idea.
     
  8. Dean

    Dean
    Well-known Member

    Joined:
    Jun 20, 2003
    Messages:
    3,034
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    106
    Location:
    London
    Ratings:
    +347
    I'll try some A/B comparisons and see, unfortunately I'm one of those people that still buy CDs rather than downloading them, I always preferred having the hard copy! I am just curious whether these prerecorded CDs would be mastered in a way to allow an improvement to be heard. They seem to be very well mastered from lounge, jazz, rock genres.

    Thanks I'll try and see, I am using a SigmaTel HD sound card which is miles better than what I was using previously on a older laptop. My main aim is getting the best playback from all the CDs I have stored onto my hard drive, I still use my Denon to play CDs of course (even using analog direct) but the convenience of having everything on a PC is too much to ignore nowadays!
     
  9. Dean

    Dean
    Well-known Member

    Joined:
    Jun 20, 2003
    Messages:
    3,034
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    106
    Location:
    London
    Ratings:
    +347

    The Kernal mixer thing was always a bad idea! Fortunately Vista does not use it anymore, see here
     
    Last edited: Feb 10, 2009
  10. Steven

    Steven
    Senior Moderator

    Joined:
    Feb 1, 2005
    Messages:
    37,620
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    166
    Ratings:
    +6,604
    XP Audio Architecture:

    [​IMG]
    [​IMG]

    Vista Audio Architecture:

    [​IMG]
    [​IMG]

    N.B. Please remove image tags when quoting
     
  11. alexs2

    alexs2
    Well-known Member

    Joined:
    Nov 13, 2002
    Messages:
    13,967
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Ratings:
    +1,671

    About time!

    I guess there were sufficient complaints about the way it handled audio to prompt some updates.
     
  12. BlueWizard

    BlueWizard
    Distinguished Member

    Joined:
    Jun 19, 2007
    Messages:
    22,352
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    166
    Ratings:
    +4,211
    Though I don't remember where I read it, I know there is a driver that replaces the XP Kernal Driver and allows full resolution output.

    I think the problem with the XP Kernal Driver is that it downgrades the sound quality to what ever it thinks is appropriate regardless of what you feed into it. So, it is always a big obstacle to sound quality.

    Let me check a few tech support sites and see if I can come up with a link to the XP replacement driver that clears up the sound quality.

    Steve/bluewizard
     
  13. reaper12

    reaper12
    Active Member

    Joined:
    Dec 4, 2006
    Messages:
    1,296
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    63
    Ratings:
    +119
    get foobar and use the kernal bypass option in the output stage, or use the asio drivers for your sound card and use that option in foobar.
     
  14. alexs2

    alexs2
    Well-known Member

    Joined:
    Nov 13, 2002
    Messages:
    13,967
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Ratings:
    +1,671
    I used the ASIO drivers for my previous PC,and the current one runs Vista.
     
  15. eaglemmoomin

    eaglemmoomin
    Active Member

    Joined:
    Jul 12, 2007
    Messages:
    2,395
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    63
    Location:
    Crawley
    Ratings:
    +226
    I believe it was more to seperate out a potentially OS killing piece of software sitting in the kernel. What it also means is that theres all sorts of DRM stylee things that you could potentially do in the audio service ie say if you used the analogue outputs of the sound card on the flip side theres loads of other things you could do. I suspect its a software engineers solution to a hardware engineer created problem (in as much that you could kill the OS with all the low level access going on potentially with memory leak problems and I guess a massive security hole too). How this sits with the ASIO stuff I do not know as I think this is only to get the PCM straight to the sound card.
     
  16. BlueWizard

    BlueWizard
    Distinguished Member

    Joined:
    Jun 19, 2007
    Messages:
    22,352
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    166
    Ratings:
    +4,211
    Let's take a quick overview of exactly when Sample Rate and Resolution mean.

    Using CDs as an example, and rounding off the sample rate to 40,000, and in this example using 20,000hz as a test frequency, how many samples does it take for a 20,000hz note? Answer: Two.

    Certainly you've all seen a sine wave, it has a rising portion mirror by an identical falling portion. A 40k sample rate will measure the voltage in two equally spaced but random locations along the sine wave. If those random sample happen to occur as the wave form in near zero, then essentially that signal voltage and signal are lost. If, by chance, it happens to sample at other locations then the resulting waveform is at best distorted.

    The same sample rate at 10,000hz, still only takes four sample voltage readings.

    Now let's look at 192Khz sample rate, and for simplicity sake we change it to an even 200,000. Now that same 20khz note will have TEN sample voltage readings taken and the waveform will be reconstructed from those 10 readings. That's not too bad, the resulting reconstructed waveform with still be slightly distorted, but will be very close to the original.

    Of course, at 10khz, it will take 20 voltage samples across the waveform, and from 20 samples we can get a very good approximation of the original signal.

    So, in a sense, this represents horizontal resolution, as the sine wave form scrolls past, we take a certain number of voltage readings, and reconstruct the waveform from that.

    Now let's look at Vertical resolution, or the actual ability to measure the voltage of any given sample. SIXTEEN bit vertical or voltage resolution allows 65,536 samples. We typically refer to this a 64k. But the voltage is measured as both plus and minus, so we can read 32,768 levels of voltage in the positive direction, and equally in the negative direction.

    To illustrate, a 100watt amp has 28 volt in both directions. So 28v/32768 samples equal a voltage resolution of 8.6x10^-4 volts or .00086 volts. Not bad, but still not close to the resolution of analog.

    However, 24 bit vertical or voltage resolution can measure voltage in 16,777,216 increments. Divided in half for plus and minus voltages, it can read in either the plus or minus direction to a resolution of 8,388,608 increments. that is extremely fine resolution.

    Using the same 100watt 28 volt example, 24 bits will read the voltage to 3.34X10^-6 or .00000334 volts, again, extremely fine resolution.

    In either case the result is something that looks like a staircase.

    In the example of a 40k sample rate and a 20khz note, it will take its first voltage reading as some random location. The voltage then jumps to that level and stays there until the next reading is taken.

    For example, if we have a +- 2 volt sine wave, and we take our first reading at 1 volt, resulting output signal will instantly jump to 1 volt and stay there until the next reading is taken. Since I've fudged the number a bit to make everything come out even, the next reading will be taken at -1 volt. Now the voltage jumps down to -1 volt and stays there until the next sample reading. But again, this is a two volt sine wave, not 1 volt, so as you will eventually see, we have amplitude error, phase error, and waveform error.

    What used to be a smooth uniform sine wave in now a sharp square wave. Further it is shifted slightly out of phase from the original signal. The digital signal output wave form will stay at is previous reading of -1 volt while the sine wave moves to -2 volts then begins to rise to zero, then it goes beyond zero to +1 volt where the next reading is take. That means that the digital waveform zero crossing point is shifted to the right relative to the zero crossing point of the sine wave. This would be easier to see if I had some illustration, but I don't.

    Now, when these signal readings are read off of the CD and converted to analog equivalents, we still have these abrupt square waves. So, we pass the signal through some filters that smooth off the sharp edges and make the signal look more like a sine wave. Though in all fairness, it is only a crude time shifted approximation of the original sine wave.

    Now I suspect when the CD standard was set, the 41k/16 bit sampling was state of the art. That was probably the best the could do consistently and reliably. So that is what we are stuck with.

    But now that the world has moved to pure digital music, meaning no physical manifestation of the music itself, meaning no CD, there is no reason why we should be stuck with this painfully low sample rate and resolution.

    Even if standard CDs can't progress because of adhering to some dated standard, there is no reason why digital music can't progress.

    If you just want background music that you are generally going to ignore, then the CD standard is probably OK. I mean we survived listening to music on AM radio for years. Even FM radio doesn't equal the standard of a vinyl recorded album, but it comes close. If you really want to seriously listen to music though, you will download your music files in the highest possible resolution and sample rate, and with the minimum amount of compression.

    Then if you want something that will fit into your MP3 player, or just some casual background music, you can down convert your master files into lesser formats.

    However, don't forget that analog music has an infinite sample rate, and near infinite voltage resolution. Nothing beats analog. At best digital approximates analog.

    Just a few random and admittedly long thoughts on the matter.

    Steve/bluewizard
     
    Last edited: Feb 12, 2009
  17. eaglemmoomin

    eaglemmoomin
    Active Member

    Joined:
    Jul 12, 2007
    Messages:
    2,395
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    63
    Location:
    Crawley
    Ratings:
    +226
    Does'nt this rather miss the point? If the audio was recorded at 44.14kHz with a 16 resolution whats the point in increasing the sample rate and bit depth when you're converting the audio back into analogue? Also I rather thought the sample rate was chosen because 20Khz is at the top end of the human auditory range and thats only for the young after the early 20s it starts to go downhill. So taking Nyquist rate into account 44.14Khz is more than enough to sample any frequency we can hear with 2 and a bit Khz of slop to prevent the aliasing you mentioned. Seems like a perfectly sound :D bit of engineering to me.
     
  18. BlueWizard

    BlueWizard
    Distinguished Member

    Joined:
    Jun 19, 2007
    Messages:
    22,352
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    166
    Ratings:
    +4,211
    First, I thought I made the very point you are making, though upon rereading, that must have been in a different thread. To YOUR point, yes, if you have low resolution, it is pointless to play it back a high resolution. You can't make something appear that isn't there.

    BUT, the original poster has all his files in the highest no compression format, so for him, the playback mode he picks is very important. I think for simply playback from his computer, he should use the playback mode that matches his files. If his files are recorded at 16b/96k, then that is what he should play them back at. That way there is no time wasted in up-converting or down-converting.

    However, for his MP3 player, he could down convert them to a more compact format if he so chose. You can always down convert, but you can't up convert.

    As to the 44.1k sample rate, sampling at 20khz might be in the extreme, but that's why I included references to 10khz. At 10khz which is a moderately high frequency that every one can hear, we only have 4 to 4.5 samples to use to reconstruct the original waveform.

    Now, I'm using simple sine waves as an example, but true music waveforms are far more complex, which means it is far more difficult to resolve them accurately with only four samples. It is just easier, to visualize sine waves. Four to five samples of a waveform, will still produce a resulting waveform that is phase shifted, with inaccurate amplitude, and a distorted waveform. In some cases, if the timing is just right, you also get frequency shift. Though with four samples the result is not as distorted as only two samples.

    To accurately sample an audio waveform with any accuracy, you need close to 10 samples per wavelength at the highest frequencies. Still, even 10 samples will produce some phase shift and measurable error. Not huge, mostly unnoticeable, but still there.

    So, for sake of argument, let's move 10 samples down to ..say...16khz, even people with diminished hearing should be able to hear 16khz as music volumes, then, unless my math fails me, find out what the sample rate with then be at 20khz.

    ..thinking...thinking...thinking...

    Pardon me for working in non-metric units.

    The wavelength of 16khz is (1,125.8 ft/s)/16k = 0.0703 feet or 0.8438 inches.

    The wavelength of 20khz is (1,125.8 ft/s)/20k =0.0562 feet or 0.675 inches.

    If we have 10 samples in 0.8348 inches, how many samples will we have in 0.675 inches?

    (0.675/0.8438) x 10 = 7.999 samples or roughly 8 samples. That represents a pretty good sample rate.

    Now, what is the sample rate that results in the above?

    ...thinking...thinking...thinking....

    7.999 samples X 20khz = 159,980 samples per second.

    Even if we cut that in half, which would only be so-so, we still have a sample rate of 79,990, which as I said, is still double the CD sample rate.

    So, bare minimum, that CD sample rate is half what it should be, and for pure near perfect fidelity, the CD rate is one fourth of what it should be.

    But again, likely when the standard was set, the specification represented the best possible technology of the day. CDs hit the market around 1982, so the standards were probably based on technology of 1980 at best.

    The Super Audio CD (SACD) was developed in 1999.

    DVD-A formats can store mono and stereo at sample rates of up to 192khz; for multi-channel formats, the best it can do is 96khz.

    With SACD it is possible to have sample rates has high as 2.8224 MHz, or EXTREMELY high. Though because it appears to be a streaming format, it seems to have a functional sample rate of about 192khz with a voltage resolution of 20 bits. (data from Wikipedia)

    So, I think that is why CD sales are down, and on-line sales are up. Partly for cost and convenience, and partly because if you want, you can get higher resolution and lower compression, and therefore better, formats in pure digital form.


    Steve/bluewizard
     
    Last edited: Feb 13, 2009
  19. Dean

    Dean
    Well-known Member

    Joined:
    Jun 20, 2003
    Messages:
    3,034
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    106
    Location:
    London
    Ratings:
    +347
    Thanks for the comments everyone, some good points bought up!
    Well I've done some initial listening tests using some very familiar CDs (albums ranging from The Cure Mixed Up to Buddha Bar!) and initial impressions are extremely interesting!

    I can change the sampling rate/bit depth output via the Vista control panal, so a/b repeat listening comparisons can be made easily. When doing comparisons from:

    2 Channel, 24 bit, 48 KHz > 2 Channel, 24 bit, 192 KHz
    I noticed a slight increase in higher frequency information and bit more immediacy to the music, but not obvious on all tracks.

    2 Channel, 16 bit, 44.1 KHz > 2 Channel, 24 bit, 192 KHz
    more pronounced difference, the 44.1 KHz sounded softer than in 24 bit 192 KHz, the opening to Love Song (Extended Mix) which has a guitar strum that noticably has more separation and slightly better decay as the notes fade out when in 24 bit/ 192 KHz. No doubt the difference in bit depth (which increase the dynamic range) as well as sampling rate applies here.

    I would say that you may only notice the difference with music you are very familar with and if you have pretty good hearing. I found a very good article on the subject HERE

    As well as explaining the similarities/differences between oversampling and upsampling, it also hints that the sine wave model, as good as a solid and mathematical model it is, should not be used solely to explain the way the human ear and brain interpretes sound and therefore music. So it seems that even though I have not the highest/purest digital from a download, sonic benefits are possible still using plain red book CDs. I will need to do some more listening this weekend!
     
    Last edited: Feb 13, 2009
  20. Dean

    Dean
    Well-known Member

    Joined:
    Jun 20, 2003
    Messages:
    3,034
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    106
    Location:
    London
    Ratings:
    +347
    Well after making listening sessions between my dedicated CDP and the Sony HTPC I decided that it is worth outputting at the highest bitrate and bitdepth when listening to music that has been ripped uncompressed. The sound always seemed to have a bit more details and just had the impression I was getting 'more of the music'. It's not perfect though as Media Centre has a tendency to crash if you skips tracks repeatedly too fast, but this is fairly minor once you know it's there and how to avoid it.

    I had only wished I tried this a few months ago as I would have avioded shelling out £230 on the Denon that I bought a few months ago, not all lost though as I did sell it on the classifieds. :devil:
     
  21. BlueWizard

    BlueWizard
    Distinguished Member

    Joined:
    Jun 19, 2007
    Messages:
    22,352
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    166
    Ratings:
    +4,211
    I'm not sure if anyone is still interested, but I found the article about the ASIO drivers for Win-XP.

    Better Music, From Your PC, for Free - PC World

    The free software is call ASIO4ALL and can be downloaded here -

    ASIO4ALL - Universal ASIO Driver

    This only matters with Win-XP, I think VISTA already has improved audio drivers.

    Just passing it along.

    Steve/bluewizard
     

Share This Page

Loading...