vivona

When I record at 16-bit, 44.1K, my Audigy 2 ZS (PCI) card limits my recorded waveform to 95% of full scale, regardless of where the input comes from. If I record at 48k or any multiple above that, I can get the recorded waveform to reach 100%. This happens regardless of recording software and regardless of whether the source is line-in, wave, MP3 or even "what you hear".

For every other soundcard I have used (SB Live, TB Santa Cruz, Intel Integrated) if I ran the recording volume too high, the clip warning on the VU meter would light and the waveform would clip at 100%. But, with the Audigy the clip warning never lights and the waveform is smashed at 95%. It makes it harder to monitor whether I am getting any clipping during recording.

I have tried the latest Creative drivers and a replacement Audigy 2 ZS card and neither helped. Moving to a 48k sampling rate immediately restores full scale recording capability. The problem only happens at 44.1K or below. This is all with the WDM drivers. I am running WinXP with all updates.

Some specialty software I use is limited to 16-bit, 44.1K, so I do not want to move to 48k. I have sent screenshots and sound files to Creative support, but so far they have not responded with a solution. In other audio forums, users have recommended the KX drivers. So, here are my questions:

1. Can anybody explain what may be causing my problem?

2. Is is feasible that the KX drivers will solve my problem?

3. Is it possible that I may have compatibility problems with the KX drivers and my recording software (Goldwave, nTrack, Sony Vegas MovieStudio, WaveRepair)?

4. If I try KX drivers and they create problems for me, does the uninstall fully remove all traces?

5. I use the IEEE1394 OHCI Compliant Firewire connector on my Audigy 2 for capturing DV from my camcorder. Will the KX drivers affect this in any way?

I look forward to your expert help. Thanks for your input!

Maddogg6

The 10Kx DSP is locked at 48Khz - So, In your case sample rate conversions are taking place.
What you see *may* be an artifact of that conversion(s). (record & playback)

I have installed different KX versions (3537, 3538h, 3538i - back to 3538h)
Durring that time I also tried YPax A4 modded drivers for A2 (NO UART1 MIDI PORT)

Using DH Driver Cleaner - does a pretty good job of removing traces of old drivers.

IEEE-1394 port - very good question - I dont have anything to plugin to it with.
I settled on 3538h instead of 3538i because of a bug in i (No Ch 10 MIDI drums)

KX handles recording differently than CL -
KX ASIO is best when used @48Khz Vs 44Khz

KX WinMM recording/playback will handle (read: do all sample conversions necessary) all the sample rates - but with winMM recording comes higher latency than ASIO.

SO - when I decided to switch to KX - I had to convert all my old projects (wav files) over to 48Khz to use KX ASIO. I suspect you'll *need* to also.

Things like media players work as they should foobar2000,winamp5, WMP10, Quicktime etc..

BUT - people with Videio Capture devices (that rely on an existing sound card to cap audio) may have problems. I have a Haupaugh WinTVUSB2 and works fine as long as I record audio from it and not my soundcard.

Messenger wont be able to Auto control Mic levels like in CL

Using Teamspeak requires KX DSP re-configuration (so you dont hear your self in your speakers when talking to others)

Im not using any of the SW you mentioned, so I cant comment on them.

Doug W

In my experience, not always 100%... but usually. However if you install the kX drivers, uninstall, then install the Creative drivers, you will not go to heaven.

vivona

After considering all the options, I feel my needs are best met if I can get the Creative drivers to work. My support case has been escalated to the Advanced Support Group at Creative. So far. they haven't been able to duplicate the problem.

What I have been able to isolate is that when recording at any sample rate lower than 48k, the recording sample rate conversion process drops the recorded waveform by .5dB which results in the peaks being about 5% lower. This problem only happens with 16-bit recordings and WDM drivers. 24-bit doesn't have the problem. I haven't tested other sample rates yet.

To illustrate the problem if I record at 16-bit, 48K using WDM drivers and increase the recording level to the point where clipping occurs, then without changing the recording level, if I record at 16-bit, 44.1k, the clipping is at 95% instead of the original 100% point. So it is the actual original 48k clipping I am seeing when I record at 44.1k. It's just 5% lower. I want 44.1k to clip at 100% (0VU) so I can take advantage of the peak indicator on my VU meter.

I had tried several versions of Creative drivers and found that one of them made the clipping point drop to 90%. So... it seems that the problem is driver related.

This forum is read by many driver experts, so I am asking for your help. It would be very useful for me to be able to isolate the exact driver file that is making the difference so I can pass that information along to Creative. The driver version that drops the clipping to 95% is 5.12.10.1162. So does 5.12.1.442. But when I tried 5.12.04.0445, the clipping was at 90%. If I can isolate the cause of the volume loss during downsampling to just one file within the driver, that would be great. Can anybody tell me which file within the driver is responsible for the WDM recording downsampling from 48k to 44.1k?

If you know the answer, please take a few seconds and reply to this message.

Maddogg6

What are you EXACT test params..

Test tone? Wave file or external signal gen - Software VU or hardware? (one that measure in '%' and not 'db'..??)

What about recording host, which one are you using - how are you measuring the recorded wave? How are you measuring ditortion? With your ears?


But...
The answer to your question should be EASILY answered by CL advanced support team - if they wont, its prolly protected by patent/license agreement/copyrights.

vivona

My test setup is to feed a 1kHz sine wave into the line-in connector using an audio generator. The Creative mixer recording level is Analog Mix 50% and the source level is Line-in 59%. Both settings represent unity gain.

Adjusting the 1kHz sine wave to approximately 3v peak ( 0 VU digital) should result in a recorded waveform at the maximum level before clipping occurs. On most recording software, that is represented on its VU meter as 0 VU. On some recording applications, the display area for the recorded waveform includes a 0 - 100% scale to measure the amplitude of the waveform in a linear fashion. Goldwave would be an example of this. 100% represents the maximum possible recorded level, or 0 VU digital. Attempting to record any higher results in digital clipping which is seen as a sharp clipping of the sine wave peaks.

To eliminate the analog line-in connector circuitry as part of the problem, I also ran the same test with the Virtual MR-1 from NTI. It creates a 1kHz sine wave that appears as a digital wave source. For that source, I set the Creative mixer wave source slider at 100%, and the recording Wave slider at 50%, both representing unity gain.

In each test I started with a recording rate of 16-bit, 48k and set the recording level to the point where the peaks of the sign wave just touched 0 VU, which is 100% of the level that is possible before clipping. Then, without changing the recording level, I changed the recording rate to 16-bit, 44.1k and the recording level dropped 5% in amplitude. Most software VU meters, that shows up as -.5dB. Since there can be a large variation in dB scales used as VU meters in recording software, and since dB levels are not linear, it is less confusing to state the drop as a 5% drop in level which is easy to estimate by eye. Regardless of how you state it, the level drops when changing from 48k to 44.1k.

When the input signal exceeds 0 VU and causes digital clipping, your recording software VU meter should indicate a full scale bar, and the peak indicator should flash, usually red, to warn you of digital clipping. That is the normal behavior I have experienced in years of working with many different PC audio setups and many different soundcards, including SB Live, SB-16, Turtle Beach SC, Intel Integrated, etc. My new Audigy 2 is the only soundcard that I have seen where digital clipping occurs at 95% of full scale on a VU meter when recording at 16-bit, 44.1k. It makes it impossible to properly monitor recording levels to guard against clipping.

As to recording software, the problem shows up exactly the same regardless of whether I use Goldwave, n-Track Studio, Sony Vegas, Creative Wave Studio or the standalone PPM peak program meter software.

Sure, Creative could tell me exactly which file could be the problem, but with all the other support issues facing them, I don't know how motivated they are to solve mine. I was hoping if I could change out just one file and make the problem go from a 5% drop to a 10% drop, I could focus them on the problem file and perhaps move them to spend the resources to fix it. Can anyone tell me for sure which driver file affects the recording SRC?

dj_stick

this is not the first instance of the card peaking at 95% - it is due to the sample rate difference

I'm afirad you'll have to either put up with it, record at 48khz, or get another card

__________________

vivona

Thanks for your reply. If the peaking at 95% is caused by the sample rate difference, why didn't I have that problem with the SB Live card in my previous PC? I recorded at 16-bit, 44.1k all the time with it and never had the problem. Didn't the SB Live card also resample to get 44.1k? The only other difference was the operating system, the old PC was Win98 and the new one is XP, but I can't see how that would make a difference.

dj_stick

the previous drivers you use may have "padded" the sound, but you'll still not get the full quality that your card is capable of at 44.1khz

__________________

vivona

Does that mean the KX drivers would not help in this case?

dj_stick

no, i don't think so sorry

the best alternative i can think of would be to record at 48khz and use a program like soundforge with a quality dithering algorithm to downsample to 44.1khz

__________________

vivona

Since I use my Audigy for IEEE-1394 connectivity and SoundFonts, I could add a second soundcard that handles 44.1k recording. I checked a borrowed Turtle Beach Santa Cruz and it did not have the 95% clipping problem. Any idea why? Does the Santa Cruz not resample?

What other decent quality modern soundcards in the $100 range handle 44.1k without resampling?

Maddogg6

Its possible the other cards are re-sampling as well, but may use a better algo in doing so. BUT Ill assume its not , and thus assume their dsp (if it even has one??) is clocked for 44Khz. I guess it architecture dependant. What parts were selected. It could very well be 44 - as back when that was made - CDs were the thing to transfer to - so then, 44khz made more sense. But now - 96 Khz is the rage - and upconverting to 96 from 48 is as simple as (X*2) conversion. (well, maybe not that simple, but mathmatically it makes more sense)

You mentioned before that your old SBlive didnt have this problem, or on-board sound, Id suggest using one of them. Or just compensate the recording level - or, use a HARDWARE mixer (Like a Behringer for ~$50US) so you can better monitor and compensate for this problem.

Welcome to real world where nothing is perfect...lol -

j/k - but we are very familiar with the term 'work-around' - and NOT because of KX - but KX IS a 'work around' for MANY other CL problems/limitations. Like how many FX can be used with Midi soundfonts - (I get 4 simultaniously) - and KX was first to offer ASIO to Live (maybe audigy too??) users.

Basically Its a matter of priority and budget. I can live with clipping at 95% (but with ASIO its not an issue as I use 48Khz)

Your problem must not have been a (priority) problem for most.
I set record levels NOT to 0db peak but -3 or even -6 anyway, and never have any problems.. WDM is not as important as ASIO to most KX users.

What 'specialty software' are you using thats limited to 16/44? maybe theres another alternative there?

vivona

I use 16-44.1k for several reasons.

I have a huge collection of song tracks that have been ripped from karaoke disks or recorded from karaoke tapes, all at 16-44.1k. I make compilations on CD to take with me when I go out singing. If I converted all to 48K, my file space would grow by 10% and all would have to be downconverted again during the CD burn. My data file backups on CD would not hold the same number of songs as I could get on an audio CD, so I would have to reduce the track count on the audio CD.

I load many of the above as tracks in nTrack and record my voice to make albums of my singing. If I converted all to 48K, and recorded my voice tracks at 48k, my file space would grow by 10% and, again, I would have to downsample back to 44.1k during the burn to CD.

I use WaveRepair, an excellent sample-level wave file repair tool to perform surgery on wave files, such as hand removing clicks, compositing vocal takes, etc. WaveRepair only works on 16-44.1k.

When the whole 95% problem started out, I was thinking that it would be a simple solution and all would be back to normal. But, noooo. Either everyone tells me to go to 48k or just record at a lower volume level. Creative Support now has me sending them extensive information about my Behringer mixer, complete msinfo32.exe exports, detailed information about my PC and every peripheral. They say their tests on the Audigy show no difference in the levels between 44.1k and 48k at 16-bit. I don't know why they want to know about my Behringer mixer since I have exactly the same problem when I record a wave playback from within the PC, not involving the Behringer or the Line-in connector in any way.

With all the effort I have spent on this, I think I would be better off just taking the loss and scrapping the Audigy, then buy a separate 1EEE-1394 PCI card for my DV input and a soundcard that works correctly at 16-44.1k and give up on soundfonts and just use whatever wavetable or software synth that comes with the soundcard. That would use up all of my PCI slots and cost me $$$, but at least I could get back to more important things. I guess a simple fix was hoping for too much.

Maddogg6

scrap the audigy and lose HW soundfont... keep it and record lower volume (then normalize??)

thats what I was saying about priority....

I converted my projects because my priority was not drive space but midi flexibility..

Good luck - and sorry if we didnt give the answer you wanted.

Did CL indicate the file that does the SRC?? - if so, mybe someone sharper than me can take that and run with it... most people here bend over backwards helping out others.. when they can..

vivona

Creative support did not indicate what file handles the SRC. When they asked about my Behringer hardware mixer it indicated to me that they are not really reading through my messages because I have several times explained that the 5% volume loss through SRC occurs even when recording a wave file being played back from the PC directly (not through the Behringer). In my last reply to them I told them that the problem was different (5% vs 10% loss) depending on which driver version I tried and gave the the driver versions. Hopefully, that will spark their attention to look at the difference between those driver versions, particularly the file(s) that affect SRC.

I suppose I could try changing each file, one by one, reboot and test and eventually find the one that drops the level. That would take a long time and I don't know if mixing file versions within the overall driver package will mess things up. There are 25 files listed as part of the driver in device manager. It sure would be helpful if someone that knew for sure which driver file handled the recording SRC for 16-bit, 44.1k. I think if I could isolate it down to one file, I could get Creative support to focus their attention in the right place.

vivona

I tried the KX drivers. They didn't help. I uninstalled KX and went back to the Audigy drivers to continue the diagnosis. Creative support continues to be confused and keeps focusing on things like my power supply and Behringer mixer. The power supply voltages and load were fine, the mixer has nothing to do with the problem - it also happens with wave files and CD playback.

It appears that the Audigy problem is a design defect that only a few users will notice. Most gamers don't do serious recording and most recordists use 16/48 or greater and the problem doesn't exist at that rate. So, to put an end to this frustration, I researched soundcards, bought an X-Fi, used Driver Cleaner to get rid of the Audigy and KX drivers, installed the X-Fi card and everything works perfectly.

Thanks for all your input. Though this problem was frustrating and time consuming, I sure added to my soundcard knowledge in the process.