Kernel Streaming/ASIO on 24bit soundcard :

2006-02-08 09:25:11

I have just bought an Audiotrak Prodigy 7.1 LT soundcard that has 24bit DACs. When i enable Kernel Streaming in fb2k then it displays : "Optimal resolution: 32 bits per sample". I would then like to know if it's allright for me to select "32 bit fixed point" under output settings ? I'm confused because under the selection box for the output bit-depth in fb2k then it says that using bit-depth higher than the soundcard supports would only degrade quality. Is it then that when selecting "32 bit fixed point", then there is still only 24 bits for data and 8 bits for padding, since if that's not the case then the 32 bits audio data would be truncated to 24 bits internally in the soundcard. And if it is true that "32 bit fixed point" only holds 24 bits of actual data, then what exactly is the difference between "32 bit fixed-point" and "24 bit fixed-point padded to 32" ? Also i have heard many people here on HA say that one should just select "32 bits fixed point" when using Otachan's ASIO plugin, and when i run his ASIO caps program to check for the properties of my soundcards ASIO drivers, then it also displays that they only support Int32LSB, so i guess that "32 bit fixed point" is right for my soundcard both in Kernel Streaming and in ASIO mode, but i would really appreciate someone to clarify for me exactly what "32 bit fixed-point" in fb2k stands for and how it differs from "24 bit fixed point padded to 32 bit", and if it's okay to use "32 bit fixed-point" on a 24 bit soundcard...

Thanks in advance.

Kernel Streaming/ASIO on 24bit soundcard :

Reply #1 – 2006-02-11 03:53:19

Could someone then please just tell me if im understanding the following right :

32 bit fixed-point :
Only 24 bits audio data and 8 bits padding in Least Significant Bits.

24 bit fixed-point padded to 32 :
Only 24 bits audio data and 8 bits padding in Most Significant Bits.

And when selecting one of them, then am i right in assuming that fb2k also informs the soundcard drivers about in what format the output is delivered in, like for example : Int32LSB or Int32LSB24.

I currently am using Kernel Streaming with the output set to "32 bit fixed-point", but i just want to be sure that no real audio data is truncated during the output transfer between fb2k and the soundcard, and also about if it would be better to use "24 bit fixed-point padded to 32" instead of "32 bit fixed-point".

By the way, i'm not diluting myself in to thinking that the perceivable soundquality will be better with Kernel Streaming or ASIO, but it's just that when i now have gotten a soundcard that dosen't resample, and since i only listen to lossless, then i just like the warm and fuzzy feeling of knowing that the playback-chain also is completely lossless(volume control and ReplayGain disabled)

Kernel Streaming/ASIO on 24bit soundcard :

Reply #2 – 2006-02-11 05:35:24

i would think that if your source files are not greater than 16 bit depth (i have yest to see any) than processing at 24 bits is not going to do anything good to the sound and will consume more processing power.

Kernel Streaming/ASIO on 24bit soundcard :

Reply #3 – 2006-02-11 07:30:27

Well I use Kernel Streaming on my laptop with my Echo Indigo DJ. The Indigo supports 24bit output so I use 24bit output. 24bit padded to 32bit could be used but AFAIK there is really no need to use it if the maximum supported bit depth for your soundcard is 24bit.

If I remember right, foobar processes files in 64bit internally, so having your output set at 24bit rather than 16bit will keep some of the processing resolution acquired when using foobar. Now I can't hear the difference between 16 and 24bit when doing this, but hey maybe someone out there can?

Kernel Streaming/ASIO on 24bit soundcard :

Reply #4 – 2006-02-11 16:48:39

@Zoom & gob

Thank you to the both of you for your replys  The reason i'm using 32 bits output instead of 24, is because the Kernel Streaming dialog in fb2k says that my soundcards optimal resolution is 32 bits.

@all

I would really appreciate if someone knowledgable in this area could explain to me the difference between "32 bit fixed-point" and "24 bit fixed-point padded to 32"...

Am i understanding the following right ? :

32 bit fixed-point :
Only 24 bits audio data and 8 bits padding in Least Significant Bits.

24 bit fixed-point padded to 32 :
Only 24 bits audio data and 8 bits padding in Most Significant Bits.

And when selecting one of them, then am i right in assuming that fb2k also informs the soundcard drivers about in what format the output is delivered in, like for example : Int32LSB or Int32LSB24.

And if Kernel Streaming says "Optimal resolution:32 bits", then does it mean the best would be for me to select "32 bit fixed-point", or "24 bit fixed-point padded to 32" ?

Thank's in advance...

Kernel Streaming/ASIO on 24bit soundcard :

Reply #5 – 2006-02-13 11:46:47

I tried to ask a similar question here short time ago. Seems those in the know are on holiday.....

Kernel Streaming/ASIO on 24bit soundcard :

Reply #6 – 2006-02-14 00:55:26

Hi Mr_Rabid_Teddybear

Yes, i am also waiting patiently, and i would really wish that one of the fb2k devs would show mercy on me, and reveal the facts about the two options

CU, Martin.

Kernel Streaming/ASIO on 24bit soundcard :

Reply #7 – 2006-02-18 04:34:47

Since i didn't get an answer to my question, then for others also wondering about this, then i have posted a quote from Peter i found in a thread :

"24bit fixed-point padded to 32bit: This produces exact same sound as plain 32bit output, since no consumer-level DAC even utilizes full dynamic range of 24bit signal. Why tell people to use advanced settings if they don't improve anything ? For placebo value maybe ? This kind of "help" is not welcome here"

And about my question about the difference between "32bit fixed-point" and "24bit fixed-point padded to 32" then i found this quote from Peter in the same thread :

"Seems you don't even know what the difference between 32bit output and 24bit padded to 32bit is either (null bits are least significant ones, not most significant)."

Source : http://www.hydrogenaudio.org/forums/index....showtopic=29626

BTW, after thinking about it some more, then i have stopped using DirectKS as output and gone back to fb2k v0.9b14's standard output, as i realize that using low-latency outputs like DirectKS/ASIO in playback programs simply dosen't make any sence as it won't give any audible improvements but only increase the chance of unstability