TheTranceFan

I'm getting a queasy feeling that this isn't possible. I'm pushing SP/DIF from a DAT at 44.1 to the optical input of my Audigy, and I want to record it, unmolested (not resampling, no efx, no filtering, no gain/attenuation), at 44.1 to hard disk as a .wav file. It's such a simple thing. Can Audigy do this, or is it gonna go to 48KHz no matter what I do?

If the answer is that I can't take a 44.1KHz SP/DIF in and record it to disk basically untouched, I'm gonna have to go get Terratec DMX 6fire or something, because I really need to do that.

Sorry if this has been asked before, I did do quite a bit of searching and didn't find this exact topic.

Thanks in advance everyone.

-Bryan
aka Geodesic

Eugene Gavrilov

this is impossible because the 10k1 and 10k2 chips operate at 48000, and even 'direct spdif recording' (a new feature of audigy2) won't help you...

however, the new p16v chip found on audigy2 can handle this
unfortunately, we cannot implement this support in our drivers simply because the chip is completely undocumented -- but we are working on this issue

btw, you could try switching your 10k2 chip to '44100' mode (a very dangerous and untested 'feature')
if you are interested to try it - let us know

/Eugene

TheTranceFan

Eugene,

Yeah, I'm definitely up to try dangerous things. BTW, I'm a programmer (worked for Apple, Adobe, etc.) and I wrote most of the audio code in Adobe Premiere up through version 6.5, so I'm pretty much game for anything.

So the clock for the EMU chip is programmable?

I'm looking forward to hearing about this

-Bryan

Eugene Gavrilov

well, the 10k2 chip supports two frequencies: 44100 and 48000
however, this is not documented and we know very little about this 'mode'
I'm afraid that re-programming the clock won't help you simply because the software layer isn't ready for this...

concerning the problem with recording unresampled spdif signal, - there might be a possibility to enhance 'direct spdif recording' feature in the future to support frequencies other than 48000...
(3533 perhaps...)

/Eugene

TheTranceFan

Well Eugene I'm all for it. I'm happy to help in any way I can. I've got a whole bunch of digital gear:

* Sony 55ES DAT (90V from Japan) - both coaxial and optical I/O
* T.C.Electronic Finalizer - coaxial I/O
* Apogee AD 500 - both coaxial and optical out
* Harmon Kardon HD 720 CD player - coaxial out
* Sound Blaster Audigy Platinum with external LiveDrive

I also have all the development tools, so if you want me to build and test stuff here I can do that. Let me know how I can help.

-Bryan (http://www.pallium.com)

Eugene Gavrilov

so, the first question: do the optical in/out work?
and how are they labelled in the mixer?
that is, which slider does function - 'coaxial' or 'optical'?
(unfortunately, we don't have any optical equipment...)

then, does 'direct spdif recording' work (48000 frequency)
(select 'generic spdif' as the source; use 'optical-in' connection)
- it should transfer digital data unmodified (16bit/stereo/48000)

then, connect your 44100-capable device to the spdif-in
now, run 'kxctrl -istat' and post the results
(it should display the spdif-in internal frequency in encoded-form)

finally, try the following: select 'direct spdif recording'; set your spdif input/outputs to 48000 and start the recording; now, switch the output frequency to 44100 (w/o stopping the recording)
what happens? (there might be a pop-up 'sync. lost' or similar... or the data might be still recording, but automatically resampled)

use 3532 rc3 for your tests...

I'm unsure, but there might be possibility to perform 'direct' recording on any frequency supported by the spdif components

/Eugene

TheTranceFan

OK, here's where I'm at. ([COLOR=red]All this with the default DSP setup in 3532rc3[/COLOR])

Question 1a: Does the optical input work: [COLOR=limegreen]Answer: Yes[/COLOR]
I can monitor the optical input by turning up the Coaxial input in the mixer's Ins and Outs page.

Question 1b: How are they labeled in the mixer? [COLOR=limegreen]Answer: Coaxial[/COLOR]
So the data shows up on the Prolog in6/in7, aka "Coaxial Left"/"Coaxial Right". No signal appears at Prolog in10/in11 ("Optical Left"/"Optical Right").

Question 2a: Does SPDIF Direct Recording (48KHz) work? [COLOR=limegreen]Answer: Apparently, BUT[/COLOR]

Setup:
* I did the test recordings using CoolEdit 2000.
* In the mixer Recording page, all inputs are muted
* In the mixer Ins 'n' Outs page, the whole bottom row of outputs are muted
* In the mixer Master page, SPDIF out frequency is 48KHz (probably doesn't matter). You said to select "Generic spdif" as the source, but that option doesn't appear in the rc3 menu for SPDIF direct recording. I see off/CD SPDIF/Coaxial/SPDIF C/I2S 0/I2S 1/I2S 2. If I select "Coaxial" from the SPDIF direct recording menu, I see the signal on CoolEdit's record monitor bars and I can record it. Here's where it gets interesting.

In CoolEdit 2000, I create a 48000/16/stereo file, and record into that. But when I listen to the file, it is apparent that the samples are really at 44100, not 48000. So if I change the sample rate (not resampling, just telling CoolEdit that the file is really 44100), the file sounds correct.

Also, there is one bit of noise at the bottom of the signal, fluctuating randomly between sample values -1 and 0. It has occurred to me that this may be a half-bit dithering signal added by the Finalizer, but I haven't confirmed that yet. The noise does not appear in the source recording (.wav), however. For the record, the material in question is pure digital, created in Reason, written out as a 44.1/16/stereo .wav, and written to CD-R with HP RecordNow.

Here's the outboard digital signal flow for the test:

* CD-R in a Harmon Kardon HD 720 CD player
= coaxial digital to =
* T.C. Electronic Finalizer
= coaxial digital to =
* Sony ES 55 DAT
= optical digtal to
* Creative Sound Blaster Audigy LiveDrive
(Yes I could have taken the coaxial digital ouptut of the finalizer directly to the Audigy.)

Here's the result of the kxctrl -istat:

[Administrator@STUDIO ~]$ kxctrl -istat
kX Driver Control Program
Copyright ⌐ Eugene Gavrilov, 2001-2003. All rights reserved.
Status:
spdif A: ffffffff 0
spdif B: 100104 703aceb
spdif C: ffffffff 0
i2s 0: 1040000
i2s 1: 1040000
i2s 2: 1040000
spdif 0: 2108504
spdif 1: 2108504
spdif 2: 2108504
spdif freq: 48000

So what's really going on? Is CoolEdit recording a sample stream that is really 44100 samples per second thinking it is 48000 samples per second? What weirds me out is that if I create a 44100/16/stereo file in CoolEdit 2000 and record into that, it too plays back too fast, and wants to have its sample rate adjusted to 40516, which is (41000 / 48000) * 44100.

That's as far as I've gotten, but it's not bad. It might be doing the right thing, albeit in a weird way.

What next? And should I be doing all this in Cubase SX (with ASIO) instead of CoolEdit 2000 (with WDM)?

-Bryan

Eugene Gavrilov

there should be no difference between WDM/Winmm and ASIO when using 'direct spdif recording'
you should check for the nature of the noise -- perhaps, playing it back via cd-digital connector is the best test [in 44100]

btw, download and install new kx.sys: http://kx.maincore.nl/kx.rar
it should lift any limitations on sampling rates
perhaps, setting your Cooledit 2000 to 44100/16/stereo will work correctly
(I'm unsure about the noise...)

/Eugene

TheTranceFan

OK I installed the kx.sys you pointed me to and now as you suggested I can create a 44.1/16/stereo track in CoolEdit and get a file that sounds right. Horray!

But it gets better!

I wanted to know whether the signal had been upsampled to 48KHz then downsampled back to 44.1KHz. Also, I wanted to find the source of the half-bit of noise I noted before. So:

* I hooked the coaxial digital out of my Harmon Kardon HD 720 CD player to the coaxial SP/DIF input of the Audigy.
* I created a 44.1KHz/16/stereo file in CoolEdit 2000, set SPDIF direct recording to Coaxial, and recorded my track directly from the CD to the Audigy at 44.1.
* The resulting file sounded fine, so I trimmed it to the exact sample start and end of the original .wav file from which the CD was made, inverted it, and mixed it with the original .wav.
* The resulting file was exactly digital zero with the exception of four samples, two groups of two consecutive samples who's values were -1 instead of zero. Those samples were probably the result of ECC by the CD player.

[COLOR=orange]Conclusion: The file I got by recording with SPDIF direct recording gave me an exact 44.1KHz copy of what was on the CD, without resampling or other molestation.[/COLOR] That's exactly what I wanted, and exactly what I got!

Further conclusion: The half-bit of noise comes from the dither module at the end of the Finalizer's output. My (oldish) Finalizer operates at 24-bit precision internally, and it even says explicitly it is dithering the signal back to 16-bit. Anyway, that noise is below -84 dB, so I'm not really worried about it.

So Eugene, [COLOR=orange]THANK YOU THANK YOU THANK YOU[/COLOR] for your help. I just donated $100US to the kXFund. Keep up the great work.

-Bryan
aka TheTranceFan
http://www.pallium.com

Eugene Gavrilov

I'm glad the recent kx.sys solved your problems -- it will be included in the final 3532 release

it seems that the 'direct spdif recording' is more power than we expected...

anyway, there's a possibility that '24bit recording' might implemented the same way -- but this requires additional testing

/Eugene

TheTranceFan

Other questions do come to mind as well...like why do the optical inputs show up in the mixer on the coaxial inputs? Are they combined by the Audigy hardware, so they're really only two different looking sockets for the same input? If not, can kX make them show up in the mixer at the proper optical inputs?

I'm happy to help figure this out, or indeed do any other testing for you that requires combinations of coaxial and optical digital, since I've got a lot of stuff here I can easily route. (I don't have any 24 bit gear here tho .)

I just pulled my tune back into CoolEdit 2000 at 44.1 out of the Finalizer. That was the problem I hoped to solve when I discovered kX on Sunday, so I'm happy as a clam over here.

Eugene Gavrilov

>> why do the optical inputs show up in the mixer on the coaxial inputs

we have based our assumptions on the idea that optical and coaxial spdif inputs are separated in hardware
due to lack of optical hardware, we've never been able to check this

so, I've posted a message in the 'general' forum to collect more information

btw, your card is 'Audigy Platinum', isn't it? pls post your hardware configuration to the above mentioned post...

/Eugene

TheTranceFan

Yes, my setup has an Audigy Platinum (but not 5.1).

TheTranceFan

Or, more specifically:

* Pentium 4 2.4GHz 533MHz FSB
* Giga-Byte 8SQ800 mobo
* 1GB DDR 3200 dual-channel memory running at DDR356
* 120 GB HD, 80 GB HD
* HP DVD-R+W, Toshiba DVD-R
* Creative SoundBlaster Audigy Platinum eX
* GeForce4 4200 128 MB DDR RAM AGP
* ATI XPert 128 PCI
* Adaptec AHA2940 SCSI Host Adapter
* 10/100 NIC
* A lot of fans

You can see my studio too, which is where this machine lives.

RIV@NVX

I think that Eugene asked for kX hardware info (kX Menu\About\Hardware info section)

__________________
-- Vedran

TheTranceFan

Ah, right. Sorry. I'm in idiot

----Hardware / Software configuration---------------
Driver Name:kX Audio Driver (Debug)
Driver Date: Apr 28 2003 23:00:10
Driver Version: 5.10.00.3532 - debug
DB Name:LiveBay1
SB0090 10k2 [d800]
PCI Information:
Device: 41102 Subsys: 511102 ChipRev: 3
Card is '10k2'
Card has MPU device
Card model is '5.1'
Card HAS AC97 codec
Codec name: TriTech TR A5/CT1297 TAT
3D Extension: No 3D Stereo Enhancement
Codec is 1.0 compliant
Capabilities[0] :
DAC resolutions : -16-bit-
ADC resolutions : -16-bit-
Ext Capabilities [0]:
Port: d800 Irq: 16
Playback buffer: 20a0
Record buffer: 4000
Number of AC3 buffers: 4
Tank memory: 256 kb
OS version: [2 5 1 Service Pack 1; 2 5 1 Service Pack 1; 156 148]