TiagoTiago

What would be the best PCI audio card that is compatible with KX that i can get? And is it worth the price or is there a cheaper one that even though it isn't top of the line it is worth to buy instead?

osm0sis

Well all these cards are decently priced since they're older, though there can be a bit of demand driving the price up at the same time since they're harder to find. I think the best card for kX is the Audigy 2 ZS and honestly I got it for pretty cheap on eBay.

See Whats The Best soundcard that work with KX project?
and kX Soundblaster Compatibility List

Happy hunting

janez

you can propably find a simple 10k1 card for free.
But my favourite is the SB0090 Audigy 1. No 24 bit recording, but a very solid card with a 10k2 Processor, lots of ins and outs and fully compatible with all the old drivebays and digi breakout cards without any cable modifications.
With a bit of patience you can get it on Ebay for a few bucks whereas the 2ZS will still cost 10 to 20 times as much.
But I must admit I donīt really know the 2ZS. What additional advantages does it have?

Russ

Yes, the Audigy and Audigy 2 ZS are both popular models.

Any 10k2 model should be good, but I would probably stay away from any "value" models or "Platinum Pro" models.

No WaveHQ, but 24 bit ASIO recording is possible with SB0090.

janez

Interesting! I thought that was the one big difference between the Audigy 1 and Audigy 2.
Can you elaborate on that? Correct me if Iīm wrong:
There is still no way to get around the 16 bit truncation whenever something enters the DSP. So 24bit recording through a digi input, if possible, wouldnīt make sense.
Do the converters, including the Aux In II on a drivebay, completely bypass the DSP?

So what would be the advantage of buying a 2ZS instead of a SB0090?

Russ

I was referring to data transfers to/from the DSP and the OS (specifically ASIO), not physical inputs (however SB0090 also supports direct recording, which does bypass the DSP).

In recent versions of kX, the data is split between two different busses in order to support 24 bit.

Honestly, I do not really know a whole lot about WaveHQ (I have never used it, as my cards do not support it), but yeah, there is still some things you can do with Audigy 2 cards that you cannot do with the Audigy 1 (related to WaveHQ (e.g. 24/96 recording), etc, and of course SB0090 is a 5.1 model, etc).

janez

Does that mean there is no 16 bit truncation when entering the DSP as there used to be? I thought that was a hardware limitation.
I guess "direct recording" means using the converters?

Iīm not even sure what WaveHQ actually is. Is it a piece of hardware on the card supported by some kx plugin or something?

Russ

Yes it is a hardware limitation, that has not changed, but it is possible to work around this limitation using multiple busses per stream.
e.g
With a 24 bit stream, each 24 bit sample can be split, such that x number of bits (below the truncation point) are sent to one bus, and the remaining bits are sent to another bus, and a plugin in the DSP takes the data from each bus and restores the 24 bit value. This is basically what is done (with 10k2 models) in recent versions of kX to enable 24 bit ASIO.

Sorry, I do not know how direct recording works internally (anything I might say would be guesswork).

Again, I do not know a lot about it, but it is related to the p16v chip on the card. WaveHQ bypasses the DSP, thus avoiding the DSP limitations (the data can be sent to the DSP via the p16v plugin, but it will be re-sampled).

There is some info here:
kX Project -- Direct Recording Guide
kX Project -- Technical Guide to Playback and Recording

Also, if you search the forum, you might (I seem to recall little bits of info here and there, but...) be able to find more info from Eugene or Max M. about these things...

janez

Very interesting!! Thanx for the info!

So does that mean the signal is split and goes through 2 different rate converters simultaneously and is then put back together?
īcause, apart from the 16 bit truncation, thats another part of the hardware limitation, isnīt it? So there probably still is no bit transparent 24 bit recording.

And how does that work in real life? I mean where is the input plugin that merges 4 FX Busses into 2. And what do I have to do on the application site... ?
I hope this is not too much of a newbie question that has been discussed extensively already. Did I miss an important thread?

As for the input ADCs on an old Audigy: I think there is no way around the DSP. The Aux2 is wired up to an I2S Input and the AC97 Mixer goes straight into the DSP. So I guess thats a unique feature of the new Audigys.

Russ

I do not know (I guess), but with ASIO we are talking about recording at 48000, so I think re-sampling is avoided there. I am not sure about the playback side, but IIRC, the resampling leaves the 1st 15 bits intact, so you can get 15 bits from each bus, giving possible bit accuracy of 30 bits(?) when using 2 busses.


I did do some testing, and it is bit accurate for the most part (other than some possible sync issues)
ASIO 24 bit bug?

The code is part of FxBusX (or ProFX:SRC) and epilog (or ProFX:ASIO). You should not have to do anything from the application side other than choosing the correct format (24/48 etc) and setting the ASIO control panel to match. (And again, I am only talking about ASIO here).

I get most of my info from the forum (and kX docs, and testing stuff), so the info is there somewhere...

janez

I donīt want to be nitpicking here, but I think the resampling is never avoided. Using 48k can just minimize the loss of Quality.


How did you do that test? Did you run some bit by bit file comparison?
Maybe weīre playing semantics here, but "bit transparent" or not is a yes or no. And due to the known hardware limitations it seems to be impossible.
So all you can get is a minimized loss. I can live with that. I think kx sounds great.

Russ

I am really not an expert, as I said, most of my info comes from info in the forum, etc.

In the following thread, Max M. says that there is no re-sampling when recording 48kHz:
Perfection Mistery ?

Also, the guide on playback and recording explains that the SRC bug (with 16/48) causes the audio data to contain a partially-wrong 16th bit (so the other 15 bits are correct). And for the new 24 bit stuff, the stream is basically split into two 16/48 streams (with bits shifted as necessary).

Regarding my tests, yes I did a binary file comparison with the recording results and the expected results (comparing only the actual audio data), and it was perfect match (other than the sync issue mentioned in the other thread) when recording multiple stereo tracks).

janez

I guess we are getting pretty offtopic in this thread. Nice chatting with you anyhow.

I can imagine that. Inside the DSP the signal is already clocked with 48 khz. So no src is necessary on the way out of the DSP. Entering the DSP is a different story....

Russ

Hehe, isn't that what I said (i.e. recording -vs- playback)?