Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Realtek ALC888S Does It Resample? (Read 13332 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Realtek ALC888S Does It Resample?

I've acquired a new PC. It came with onboard ALC888S sound. I've also installed my old soundcard (Terratech EWX24/96) and a friend has given me his old Creative Audigy 2ZS.

Having 4 channels of input available appeals to me so I would like to use the ALC888S or the Audigy as well as my Terratech. Now, I know Creative cards resample everything to 48khZ and that their resampling doesn't have a good reputation but I'm not sure if the ALC888S does too. The data sheet says it supports 44.1, 48, 88.2, 96 and 192 khZ and doesn't mention resampling.

Does anybody have a definitive answer on this?

Realtek ALC888S Does It Resample?

Reply #1 –
I'm pretty sure that all HDA codecs have an oversampling DAC, so yes, they resample.
They can output audio on all jacks and therefore need very simple reconstruction lowpass filters. This would not be possible otherwise.

Realtek ALC888S Does It Resample?

Reply #2 –
Quote from: soulsearchingsun on
I'm pretty sure that all HDA codecs have an oversampling DAC, so yes, they resample.


Oversampling and resampling refer to different things in this context.  The OP is asking if his sound card resamples everything to 48kHz or if it can be reclocked to different sampling rates.  Its entirely possible to build a DAC that can use multiple sample rates and also oversample.

Realtek ALC888S Does It Resample?

Reply #3 –
My previous motherboard had a alc889a onboard which was bitperfect at 44.1khz. Not quite the alc888s, but i would not be surprised if it's very similar.

My currect motherboard has a VIA VT1708S which unfortunately is unable to output 44.1khz via digital out. So i let foobar resample everything to 48khz.

Realtek ALC888S Does It Resample?

Reply #4 –
Quote from: botface on
I've acquired a new PC. It came with onboard ALC888S sound. I've also installed my old sound card (Terratech EWX24/96) and a friend has given me his old Creative Audigy 2ZS.

Having 4 channels of input available appeals to me so I would like to use the ALC888S or the Audigy as well as my Terratech.


If you are trying to do 4-channel recording or playback using two stereo audio interfaces, your results are going to be suboptimal, since each audio interface has its own clock. This means that the channels aren't going to be in precise synchronization with each other. Each pair of channels may start and stop at slightly different times when you start a recording or playback operation. They won't stay in precise synch in steady use.

Quote
Now, I know Creative cards resample everything to 48khZ and that their resampling doesn't have a good reputation


Depends which Creative cards you are talking about. The resampling problem was pretty egregious in the early SB Live! cards, but that was addressed in later versions. The SB Audigy cards also resampled, but the audible problems due to resampling were pretty well addressed in them.

Quote
but I'm not sure if the ALC888S does too. The data sheet says it supports 44.1, 48, 88.2, 96 and 192 kHz and doesn't mention resampling.


My general rule is that you should judge audio products based on what they do, not how they do it. If you have an concerns about the fidelity of a given audio interface, it can be tested quite simply using the Audio Rightmark program, which is a free download and requires only  simple jumper cable (s)  to use.  Or surf the web for an published Rightmark test of the audio interface you have before you.

Realtek ALC888S Does It Resample?

Reply #5 –
Quote
If you are trying to do 4-channel recording or playback using two stereo audio interfaces, your results are going to be suboptimal, since each audio interface has its own clock. This means that the channels aren't going to be in precise synchronization with each other. Each pair of channels may start and stop at slightly different times when you start a recording or playback operation. They won't stay in precise synch in steady use.


Presumably the same applies if I record a couple of tracks on one soundcard and then use the 2nd one for monitoring while I record a couple more tracks on the 1st one?

Quote
My general rule is that you should judge audio products based on what they do, not how they do it.


Mine too. My understanding is that Creative's resampler introduces HF noise. These days I can't hear above 12.5khZ so I'm not sure I'd be able to judge.

Realtek ALC888S Does It Resample?

Reply #6 –
Quote from: botface on
Quote
If you are trying to do 4-channel recording or playback using two stereo audio interfaces, your results are going to be suboptimal, since each audio interface has its own clock. This means that the channels aren't going to be in precise synchronization with each other. Each pair of channels may start and stop at slightly different times when you start a recording or playback operation. They won't stay in precise synch in steady use.


Presumably the same applies if I record a couple of tracks on one soundcard and then use the 2nd one for monitoring while I record a couple more tracks on the 1st one?

Yes.  Of you want 4 tracks in near-perfect synch, then you have to make them at the same time using the same audio interface. The individual channels in most audio interaces are usually really well-matched.  The analog input buffers are often enough better that theiy don't contribute signficiant noise, and the converters are all identical sigma/delta converters whose noise floor is strongly influenced by the dither whose source is in the digital domain. I undertand that the better multichannel ADC chips offset the dither for the various channels so that the noise sums geometrically, not linearly if you sum the channels.

Quote
My general rule is that you should judge audio products based on what they do, not how they do it.


Mine too. My understanding is that Creative's resampler introduces HF noise. These days I can't hear above 12.5khZ so I'm not sure I'd be able to judge.


If there is HF noise in their resampler, then this suggests that they are using shaped dither, which is a pretty sophisticated move on their part for a consumer audio interface. As usual it is all about quanitifcation.  OK, the noise is there. How much noise is it? 

Realtek ALC888S Does It Resample?

Reply #7 –
Well, I found some RMAA results and it seems the noise level is close to -96DB (A). So it seems they had sorted out the resampling problem with the Audigy cards as you said or maybe it handles 16/44 natively. Either way it seems I don't have to worry about noise from this card

Realtek ALC888S Does It Resample?

Reply #8 –
I have an older Audigy card and can tell you that HF sweeps and udial have very obvious audible problems @ 44.1kHz.  A recent reply to one of my posts also suggests that the Audigy2 has problems as well, so no I don't think Creative has "pretty well addressed" the resampling issue with these two particular lines of cards whatsoever.

Realtek ALC888S Does It Resample?

Reply #9 –
Thanks greynol. It's curious that the RMAA noise figures look so good then (www.vegalab.ru/forum/attachment.php?attachmentid=103511&d...).

As I said, my worry is that I simply won't detect any noise due to my limited HF hearing and as I'm sometimes "engineering" for other people I wouldn't want to give them something that sounds OK to me but is actually unacceptably noisy. I'd like to use the Audigy as it's the platinum version and has some handy inputs on the front panel. I guess the sensible thing to do would be to simply record everyting at 48khZ and resample the end result to 44.1

Realtek ALC888S Does It Resample?

Reply #10 –
The test I did on my Audigy with sweeps and udial didn't create HF noise (at least not for me) but obvious aliasing artifacts in the mid-range. However it's hard to notice in real music.

BTW: My onboard ALC888 let's me select the desired output rate in the driver. It resamples everything else to that.
Here's a screenshot (don't mind the digital, it does so for analog out too).

Realtek ALC888S Does It Resample?

Reply #11 –
Indeed, the problem was aliasing which I am fairly certain falls under the umbrella of what Arnold was talking about when he said, "The resampling problem was pretty egregious in the early SB Live! cards, but that was addressed in later versions. The SB Audigy cards also resampled, but the audible problems due to resampling were pretty well addressed in them."

Realtek ALC888S Does It Resample?

Reply #12 –
I don't think the RMAA tests are particularly sensitive to aliasing.

Realtek ALC888S Does It Resample?

Reply #13 –
If I recall corectly, resampling didn't fix the aliasing issue. I had to add a 19khz lowpass to the playback output. Not that easy to pull off if you intend to use it for recording.

Yet the Audigy allows you to use the KX driver and it's powerfull DSP (take a look at 2 & 3).

Realtek ALC888S Does It Resample?

Reply #15 –
Quote from: saratoga on
I don't think the RMAA tests are particularly sensitive to aliasing.


It just takes a little fiddling. Let's say that I have a 16/44 sound card that I want to test with RMAA and include testing for aliasing.

I create a 24/96  test file with RMAA, and play it on a good 24/96 audio interface,  recording it at 16/44 using the 16/44 audio interface I want to test. I then use a good upsampler to turn the file I recorded at 16/44 back into a 24/96 file. I then use RMAA to analyze the 24/96 file, paying particular attention to response in the 16/44 transition band around 22 KHz.

Realtek ALC888S Does It Resample?

Reply #16 –
How does testing the ADC tell me anything about the card's DAC capabilities regarding aliasing?

Realtek ALC888S Does It Resample?

Reply #17 –
Quote from: googlebot on
How does testing the ADC tell me anything about the card's DAC capabilities regarding aliasing?


DACs can't alias.  A digital data stream cannot contain digital data representing any frequencies hitgher than Nyquist.  They can produce spurious responses. If you want to look at the spurious responses of a DAC with RMAA capture the output, you have to use a slightly different work flow.

This time create a 24/96 test .wav file with RMAA and downsample it to 44/16 with a good resampler. Run the RMAA test and capture the output of the DAC with a good 24/96 ADC. When you process the RMAA data, you will find your spurious responses.

An exmple of an audio interface that is good enough for testing 16/44 ADCs amd DACs would be M-Audio's AP 24192 running under XP. Last time I tried to loop my AP 24192 under Windows 7 in 64 bit mode, its performance was about 10 dB worse than it was under XP and basically looked like it was running in 16 bit mode.

Realtek ALC888S Does It Resample?

Reply #19 –
Quote from: Arnold B. Krueger on
A digital data stream cannot contain digital data representing any frequencies hitgher than Nyquist.

According to what I've learned the spectrum of digital data is periodic with the sampling frequency. The spectrum is weighed with the sinc-function in real systems. That's the reason you need an reconstruction lowpass filter to avoid what I would call aliasing.
So yes, digital data doesn't contain different data above Nyquist.

Is that point of view wrong?

Realtek ALC888S Does It Resample?

Reply #20 –
Quote from: soulsearchingsun on
Quote from: Arnold B. Krueger on
A digital data stream cannot contain digital data representing any frequencies higher than Nyquist.

According to what I've learned the spectrum of digital data is periodic with the sampling frequency.


Not true. Any spectral limitations on sampled data are related to the period of time over which samples were taken.

A more proper statement would be that the spectrum of a measured signal is at least periodic within the total measurement time window It could have a longer period, but you would have to make some assumptions due to lack of sufficient data to know for sure.

This applies to both analog and digital signals. Sampled data can be either analog or digital. Sample-and-hold circuits predate the widespread use of digitized signals by several decades.

An audio engineers inside joke about CD players is that the lowest frequency that they can play is something like 1/4800 of a Hz.  There would be some ambiguity about signals that are less than 1/4800 of a Hz apart

Modern Hard drives enable a considerably lower low frequency bandpass than CDs as well as better resolution of  frequencies that are close to each other. Of course, you'll never hear the difference.




Realtek ALC888S Does It Resample?

Reply #21 –
Quote from: greynol on
Before proceeding any further with this, have you at least listened to the 44.1kHz udial sample on either an Audigy or Audigy 2?


I did 19 & 20 KHz two tone tests on an original Audigy some years ago, and found that nonlinear distortion was something like  66 dB down (less than 0.1%). Therefore, if you do a UDial test and crank 'er up, you'll probably hear something. At normal gain settings, not so much.  You can reset the frequencies that RMAA uses for IM testing and collect information about high frequency IM, but again the source could be anything in the card's signal path.

One problem with the UDial test is that it does not provide accurate, reliable quantification. It sheds no reliable light on resampling problems in audio interfaces, since any spurious responses that someone might hear might have some other cause than poorly done resampling. The problem could even be other components in the signal chain, including amplifiers and transducers.

Realtek ALC888S Does It Resample?

Reply #22 –
A simple "no" would have sufficed.

Had you listened to udial played back at its native sample rate and then played it back again converted to 48kHz using an SRC known to perform well (the one in foobar2000 will do just fine), you would have saved yourself (and us) from making unnecessary speculative remarks.

As an end user, I don't care what part of the Audigy is causing the problem.  I simply know that it resides in the card itself and fixing it for my own personal needs is quite trivial.

Realtek ALC888S Does It Resample?

Reply #23 –
Quote from: Arnold B. Krueger on
DACs can't alias.

Why not? As I've learned from you, contemporary DACs up-sample in the digitally domain in order to improve the reconstruction filter. Up-sampling is equivalent to some kind of interpolation which in general creates frequencies above the Nyquist frequency (if not taking into account all samples from minus until plus infinity).

Maybe it's even more worth when reconstruction is done completely in the analog domain.

Realtek ALC888S Does It Resample?

Reply #24 –
Quote from: greynol on
A simple "no" would have sufficed.


It would have been insufficient to address the relevant issues.

Quote
Had you listened to udial played back at its native sample rate and then played it back again converted to 48kHz using an SRC known to perform well (the one in foobar2000 will do just fine), you would have saved yourself (and us) from making unnecessary speculative remarks.


That I have not done the above is your speculation. I have indeed done so. No surprises. 

I have also applied nonlinear distortion to UDial to see what it is supposed to sound like when there is too much nonlinear distortion. I didn't happen to have anything on hand that messes  up, including the ADI 1886 on-board interface in this PC and the Realtek ALC 880 on-board interface in the PC across the room.

I did however pull an Audigy from my sound card archives and installed it with the latest-greatest CL drivers for XP in a specially built clean-install machine. Udual drives it crazy.

Quote
As an end user, I don't care what part of the Audigy is causing the problem.  I simply know that it resides in the card itself and fixing it for my own personal needs is quite trivial.


Interestingly enough, the Audigy appears to be a poster boy for the benefits of avoiding making files that peak in the last dB below FS.  If I attenuate UDIAL by 0.5 dB its artifacts are far less audible, and if I attenuate it by 1 dB, they pretty much go away.  The rarity of files that are composed of  tones > 19 KHz that peak  within 1 dB of FS explains why people don't notice this problem except with arcane test files..

Just a friendly reminder - this thread is about resampling and the ALC888S, not the Audigy.  Please explain how the poor performance of the Audigy sheds light on either issue. The good performance of the ALC 880 suggests that the ALC 888 might perform as well.  Even the Audigy lacks severe problems with Udial if its peak level is reduced below -1dB F.S