|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
66.215.103.31
I have a pc based system ... right-now running from the source card on my ibook. The more I read about USB DACs, the more I'm confused. Is I2S better ? Whats with 192kHz and 44.1 kHz. Can the DAC read 192 from the hard drive? or is it just at a maximum 44.1 ... then oversampled to 192?
Last but not least .... is it better to go for a usb source like Trends UD10 or better a DAC?
Thx
Follow Ups:
Oh boy .... that was eye opener for me
Thanks
Presto and GR!!!!
Greetings:Is I2S better ?
I2S is the "native language" of modern DAC chips. I2S by itself is not "better" per se, but what is better is going directly from data (USB) to I2S without going from data -> SPDIF -> I2S. Losing the extra conversion step is supposed to be a good thing as far as jitter is concerned. You can't compare I2S to SPDIF because (AFAIK) I2S is only good for board level distances, whereas SPDIF is good for transmission over cables. They're entirely different specifications.
Whats with 192kHz and 44.1 kHz.
These are sample rates. There are two important numbers in digital audio - sample rate and bit depth. The actual data transmission rate (for SPDIF) is # channels x bitdepth x samplerate.
So, for example, stereo CD files are
2 channels x 16 bits x 44.1 khz = 1411 kb/s = 1.411 Mb/s
which is the data transfer rate for Redbook CD.LPCM files from video discs are commonly at 48khz sample rate.
2 channels x 16 bits x 48 khz = 1536 kb/s = 1.411 Mb/sThen you get into 24 bit hi-res stuff...
2 channels x 24 bits x 48.00 khz = 2304 kb/s = 2.304 Mb/s
2 channels x 24 bits x 88.20 khz = 4223 kb/s = 4.223 Mb/s
2 channels x 24 bits x 96.00 khz = 4608 kb/s = 4.608 Mb/s
2 channels x 24 bits x 174.0 khz = 8447 kb/s = 8.447 Mb/s
2 channels x 24 bits x 192.0 khz = 9216 kb/s = 9.216 Mb/sNotice that DVD-V with a maximum transfer rate of 10 Mb/s would need to double just to handle stereo 24/192 audio material. This is why we have DVD-A discs - they use all that bandwidth for audio with no video.
In order to have audio with video, you need to get into multichannel lossless compression techniques like those offered by Meridian. "Meridian Lossless Packing" (MLP) allows up to 6 channels of 24/96 of 24/192 material to be compressed into a single data stream that is passable over a "single-wire" consumer SPDIF interface (maximum 4608 kb/s). "Dual-wire" versions of SPDIF or AES are to handle 24/192 material.
Can the DAC read 192 from the hard drive?
Depends on the DAC. You need to read the manufacturers specifications for every DAC to see which sample rates are supported. In most cases, you cannot run 192khz signals over consumer single wire SPDIF. DVD-A material at 24/96 and 24/192 is downmixed to 16/48, so piping hi-res material over digital lines into a DAC is not possible without doing serious (and potentially illegal) modifications to playback hardware. Even software DVD-A players are programmed to downsample copy-protected material. The only time you can play back hi-res material over digital lines is to get into hi-res DVD-V material such as some of the Chesky records recordings and AIX records for example. DVD-V does not use DVD-A watermarking envcryption, and allows the full resolution digital signal to be tranmitted with no downsampling IF the player and DAC are capable of handling the sample rate.
or is it just at a maximum 44.1 ... then oversampled to 192?Most DACs oversample to a multiple of the original sample rate - except, of course, non-oversampling DACs. Now, with true OVERsampling, you could get 2x44.1 = 88.2khz and 4x= 174khz. In order to get 192, you need to perform resampling. SPDIF encoders like the CS84xx family resample the output. For example, the Behringer DCX 2496 has a CS8420 SPDIF encoder on the input and it resamples EVERY THING from 44.1 to 96khz to 96kz.
Just because a DAC does oversampling does not mean the resolution of your source material is increased. This is not why it's done. It's done to move what is called the "Nyquist Frequency" up, which is defined by samplerate/2, or 22.05khz for 44.1, 44.1khz for 88.2 and 88.2khz for 174.4. Above the nyquist frequency, a filter is required to filter out artifacts of the D/A process. Circuit designers and audiophiles alike tend to believe that pushing the frequency of this filter upwards (using oversampling) is a good thing. Others believe there is a sonic penalty for oversampling and prefer to use a special non-oversampling DAC.
Last but not least .... is it better to go for a usb source like Trends UD10 or better a DAC?The trouble with comparing USB capable audio equipment across the board is that there seems to be no "USB specification" for audio data transmission. This means that those who implement USB connectivity are free to use USB drivers from Windows XP, Windows Vista, MAC, or write their own special drivers. They are also free to use different USB chips. The latter issue is probably the greatest concern. Some USB chips spit out SPDIF, which has to be converted to I2S to "talk" to the DAC. Some USB chips spit out I2S and can talk to the DACs directly. This concept is the latest buzz in the computer audio asylum. Unfortunately, some of us are using digital signal processing (DSP) to perform active crossover functions, and need 6 or 8 channels and not just 2 - so for us, USB solutions are not quite there yet since they typically only have 2 channels.
Hope this helped.
Oversampling is required as a means of implementing a digital filter in order to bandlimit the signal to the Nyquist frequency to prevent aliasing. The Nyquist frequency is solely a result of the original sample rate. Oversampling does, as a consequence, push the D/A conversion artifacts (sample glitches) to a higher frequency which allows the use of a more gentle filter on the DAC output (analog domain) for their removal.Upsampling does change the Nyquist frequency to a degree, or at least spreads the spectrum and allows an anti-alias filter with a more gentle slope. Or something to that effect :-)
Slider:Okay - if oversampling does not do it...
So what if you upsample first before you get to your DAC. Is not the Nyquist frequency pushed up in THAT case?
I thought if you upsampled your original sample rate now *is* the new sample rate...
Just wondering.
Cheers,
Presto
Upsampling is kind of like resampling the signal at a higher rate, as you say in your message. If it could be done with a high enough precision, it would seem to be almost a "free" process to increase the sample rate and avoid the problems that result from extremely steep filtering at the Nyquist frequency. Unfortunately, it seems to just add a new set of problems, along with the additional filter stage. Guess nothing comes for free in the real world. But the tradeoffs may result in something better.
Slider:I guess that's why some like upsampling and others don't!!
Thanks for chiming in...
nt
Alan:I've learned so much interacting with so many people on so many different subjects - it's really amazing.
I can help others a little bit now, and when I am wrong, others chime in and keep me on the straight and narrow path towards audio enlightenment! ;)
Audioasylum is indeed a marvelous little planet. :D
That April Fool's joke almost gave me a stroke...
Confused,USB 1.1 is maxed out at 12MHZ bits per second. In essesence that would allow output to some dacs at 24/192. Most of the commercial USB stuff is 24/96. Some of the driver required hardware is capable of 24/192.
Upsampling to me is evil as is digital filters and oversampling. These all fall into the same category in my book as they are all typical digital filters. My feeling is that ounce touched by these the essence of the music is lost.
I prefer to preserve the data as best I can between the computer and the dac. For that reason 44.1 is fine for me. The lossless compression means bit true data.
Any higher formats are basically going to mean allot more storage. You can do the math but go to 24 bits and expect the size to double go to 96K and then even more.
equation data size = 2(stereo)x BITS x samples per second x song lentgth in seconds.
Yes I2S is better as some of these are going USB-> SPDIF then into an SPDIF receiver and that to me is a waste of time. The idea is to get away from SPDIF.
| ||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: