Custom Fxbus, Prolog, Epilog for Musicians

Lex Nahumury · Nov 2, 2003, 03:27 PM

mAlo,
>>I mean the ability to select in the combos of the Eplt even the asio outputs

I see Well, the 'Eplg' which I released in this thread allready does that.
But the design problem with having selectable ASIO channels in a 4x2 config,
is that *unused* asio channels have to be *muted* to "empty" the asio buffers with zeros.
This waistes ~14 extra dsp instructions, and ~2 gprs!
So we loose some of the resource saving advantage,
and it is not possable to load more then 1 'Eplg' because of this asio *muting* feature.
(read my correspondance with Travelrec)

So from a 'resource saving' point of view, there was no profit and I decided to
make the 'Asio' plugin instead, wich offers all asio channels at the same resource "price".

To conlude;
Almost everything is possable.
I could, for example write one single Asio channel pair plugin wich uses only 2 instr. and 2 gprs.
This would be great for saving resources if you only have to record to 1 asio channel pair.
But I'm afraid that releasing such plugins would confuse kX users who have little kX DSP knowledge
thereby raising the nr of "how to" questions in general.

Feel free to ask any questions, suggestions etc.
That is what this thread is meant for.

/LeMury

mAlo · Nov 2, 2003, 07:08 PM

Quote:

Originally posted by LeMury

I could, for example write one single Asio channel pair plugin wich uses only 2 instr. and 2 gprs.

Do something similar, please

Quote:

Originally posted by LeMury

But I'm afraid that releasing such plugins would confuse kX users who have little kX DSP knowledge
thereby raising the nr of "how to" questions in general.

I'm agree with you, but I do not see why an unexperinced user would have to use similar
plug-ins

Quote:

Originally posted by LeMury

Feel free to ask any questions, suggestions etc.
That is what this thread is meant for.

/LeMury

Thanks a lot for your patience!
And for your being here

NattyDredd · Nov 10, 2003, 12:11 PM

it makes more sence, especially as Im a newbie to the DSP, I'd be v. interested in those there things...

K8L · Nov 10, 2003, 06:14 PM

Is there a way to download these drivers ? I did it once, i can't find 'em anymore.

Thanx in advance.

Lex Nahumury · Nov 10, 2003, 06:32 PM

Quote:

Originally posted by K8L
Is there a way to download these drivers ? I did it once, i can't find 'em anymore.

Preliminary versions of Src, k1lt, k2lt and Asio are bundled with the latest 3536c dev. release;
http://www.hardwareheaven.com/showthre...threadid=30483

/LeMury

K8L · Nov 10, 2003, 06:43 PM

Quote:

Originally posted by LeMury
Preliminary versions of Src, k1lt, k2lt and Asio are bundled with the latest 3536c dev. release;
http://www.hardwareheaven.com/showthre...threadid=30483

/LeMury

Thanx man. It's great job.

kokoon · Nov 12, 2003, 09:31 PM

pretty please, point me to what i have to downloat for the src and the other plugins. for kx version 35.

Lex Nahumury · Nov 12, 2003, 11:29 PM

>>point me to what i have to downloat for the src and the other plugins. for kx version 35.

Sry, there are no downloads for 3535 anymore since they got bundled with 3536c driver release.

/LeMury

kokoon · Nov 13, 2003, 07:47 PM

hmm... what if i dont want 3536?

Daniel Drummond · Nov 13, 2003, 08:04 PM

Then you won't have LeMury's plugins... hehehe

Lex Nahumury · Nov 13, 2003, 08:55 PM

>>hmm... what if i dont want 3536?

I see. Well, it turned out that a lot of (new) users don't understand how an
"open" project like kX works. I.e. they simply assume that every (not included) dsp plugin
should automaticly work with every driver release.
If it doesn't work ppl started reporting 'driver bugs' and stuff like that.
To avoid such confusion there are no more "unsupported" plugin versions pubicly available for download.

However, since this obviously doesn't apply to you (as a 'kX veteran') send me an e-mail,
and I'll send you the 'old' versions, or contact Daniel Drummond for a copy.
Note that I will not maintain any updates nor support for these old versions.
All tech-info can be found in this thread.

/LeMury

Daniel Drummond · Nov 13, 2003, 11:54 PM

Yes Kokoon, I can send those to you if you want. Just let me know!

kokoon · Nov 16, 2003, 05:19 PM

well since version 3536 looks pretty stable by now, i think i'll just try it out. thanks anyway!

eyagos · Nov 21, 2003, 08:14 AM

Hi.

There is a bug in the plugin src included in 3536c: The Line In has a lower level than epilogs one (-12dB). I have been looking, and when any AC97 source is selected, the output is multiplyed by 0.25

Regards.

Lex Nahumury · Nov 21, 2003, 08:50 AM

>>There is a bug in the plugin src included in 3536c: The Line In has a lower level than
>> epilogs one (-12dB). I have been looking, and when any AC97 source is selected, the output is >>multiplyed by 0.25

First;
You probably mean Prolog instead of Epilog, since Epilog does not do AC97 input at all.

Second;
Src's AC97 inputs uses NO digital attenuation/amplification like the new Prolog does!!!
This is intentional to give better SNR. (the 0x20000000 mult is standard)
Since 3535 now you can get ~-75dB noise floor instead of ~-63dB.
(please read recent threads about ac97 input levels, SNR etc)

I hope it's clear now. (nevertheless I will check it)

Regards,

/LeMury

eyagos · Nov 21, 2003, 12:30 PM

First:
I not only have readed the recent threads, but I have checked too that multiplying by 4 were incorrect. I normally feed into the Line In the signal from a TV card and I got a maximum level of 0dB. But feedind in it the HI-FI, wich can provide more voltage, I can get +12dB, so the plugin is correct.

Now I ask myself why in 3536c (the version I'm ussing now) prolog all AC97 inputs are multiplyed by 0x2000000, except Line In wich is multiplyed by 0x7FFFFFFF.

And other thing, although I understand you, you are not talking well when you say that SNR is increased by multiplying by 0.25, because you are reducing noise and signal at the same amount. What you must say is that you are increasing the signal (and noise) range playable in the speakers.

And second:
Of course I wanted to say 'epilog' when I sayd 'epilog'. Sorry, prolog

Regards.

Lex Nahumury · Nov 21, 2003, 01:30 PM

>>you are not talking well when you say that SNR is increased by multiplying by 0.25,
Sorry Eyagos, but that is not what I said.

The SNR of 'kX's ac97's ADC to DSP signal path' *has* increased since 3535
because there was an 'invalid' (digital) amplification *after* ADC stage.

Eugene/Max please FIXME here..!

Sorry Eyagos, I'm just getting tired of the subject since I already posted
many posts with extensive details as did others.

Regards,

/LeMury

eyagos · Nov 21, 2003, 03:53 PM

Don't bother what you have concluded in other posts with other people, but SNR has not been increased at all:

My TV card with 'old' drivers -> Floor = -60dB = 0.001 // Maximum = 0dB = 1 // SNR = 60dB
My TV card with 'new' drivers -> Floor = -72dB = 0.00025 // Maximum = -12dB = 0.25 // SNR = 60dB

If we are talking about 'the same SNR' ...

Lex Nahumury · Nov 21, 2003, 04:50 PM

>>My TV card with 'old' drivers -> Floor = -60dB = 0.001 // Maximum = 0dB = 1 // SNR = 60dB
>>My TV card with 'new' drivers -> Floor = -72dB = 0.00025 // Maximum = -12dB = 0.25 // SNR = 60dB

Yep, the above is correct.

Only,..The difference now is that you can Increase the *analog* input signal to get to the same 0dB,
but *without* digitally amplifying *after* A/D conversion.

I.E. If you would take a pre-amp and amplify your TV-card's analog signal so that it uses ADC's full range
you will get 0dB again..!
But this time without digitaly amplifying ac97's noise floor..!

So the SNR of AC97 input in kX has been improved.

In your case, if you leave the analog TV-card signal the same, there is NO SNR improvement
because you will (again) digitaly amplify just like it was before 3535.

In (yet) other words;
We do not have to digitaly amplify ac97's noise floor if we make sure our analog input signal
is high enough to get 0dB.

/LeMury

eyagos · Nov 21, 2003, 05:55 PM

yes yes yes. We say the same:

Now we can achieve a -72dB floor noise. Suposing that these are the minimum that can be achieved with an Audigy2 (mine), and knowing that 'SigmaTel STAC9723' (mine) gives -90dB SNR, we can say that the Audigy introduces 18dB of aditional noise.

Then, let's say that Audigy2's AC97 inputs (mine) have a -72dB SNR. Okey, but drivers can't increase/decrease this SNR!!!!!!! It's only that yesterday these SNR dBs where distributed between [-60dB, +12dB] (and then loosing that 12dB becuse of clipping) , and today are distributed between [-72dB, 0dB] (what is more correct).

But the noise HEARD inside a 'clean' signal is the same today than yesterday or, in other words, the heard sound has the same quality, or the same SNR (think in a -20dB clean signal). Do you understand what I mean now?.

Lex Nahumury · Nov 21, 2003, 07:34 PM

>>Then, let's say that Audigy2's AC97 inputs (mine) have a -72dB SNR.
>> Okey, but drivers can't increase/decrease this SNR!!!!!!!

No, of course not. This is defined in hardware.

>>It's only that yesterday these SNR dBs where distributed between [-60dB, +12dB]
>> (and then loosing that 12dB becuse of clipping)
>> and today are distributed between [-72dB, 0dB] (what is more correct).

"Today" the analog Input signal, relative to ac97's noise floor, can be set ~10db higher without clipping.
To me that translates as an SNR improvement. (or SNR correction, if you wish)
I have no other word for it.

But realy Eyagos,.. we probably mean/say the same but only interprete it differently.
I understood your point but if you don't mind, let's drop this discussion.
All I'm interested here in this thread is to make plugins that are within kX's I/O specs.

Regards,

/LeMury

Max M. · Nov 21, 2003, 09:03 PM

>But realy Eyagos,.. we probably mean/say the same but only interprete it differently.

yep, it seems like that.
Although it would be more correct to say that "older versions just forced user to choose not correct settings with worse SNR" (and user can get the same signal path as it has now by setting "ac97" fader to 25%)...
E.g. newer driver versions do not improve SNR itself but "force the user to deal with initially better setting which gives better SNR"... something like that...

>knowing that 'SigmaTel STAC9723' (mine) gives -90dB SNR, we can say that the Audigy introduces 18dB of aditional noise.

Well, i'd said that these values written in codec specification are more like "abstract values", like "the best expected SNR in ideal enviroment" (when chip is not in machine and when it is not connected to anywhere at all)...

"that the Audigy introduces 18dB of aditional noise" - hmm, no, i would not say that...

Nov 10, 2003, 12:11 PM	#153
NattyDredd DriverHeaven Newbie Join Date: Nov 2003 Location: UK Posts: 4 Rep Power: 0	Sounds like a fine plan it makes more sence, especially as Im a newbie to the DSP, I'd be v. interested in those there things...

Nov 10, 2003, 06:14 PM	#154
K8L DriverHeaven Newbie Join Date: Oct 2003 Posts: 2 Rep Power: 0	Avaibility ? Is there a way to download these drivers ? I did it once, i can't find 'em anymore. Thanx in advance.

Nov 12, 2003, 09:31 PM	#157
kokoon DriverHeaven Addict Join Date: Feb 2003 Location: slovenia Posts: 269 Rep Power: 0	pretty please, point me to what i have to downloat for the src and the other plugins. for kx version 35. __________________ [kokoon]

Nov 13, 2003, 07:47 PM	#159
kokoon DriverHeaven Addict Join Date: Feb 2003 Location: slovenia Posts: 269 Rep Power: 0	hmm... what if i dont want 3536? __________________ [kokoon]

Nov 13, 2003, 08:04 PM	#160
Daniel Drummond DriverHeaven Senior Member Join Date: Dec 2002 Location: Brazil Posts: 1,137 Rep Power: 0	Then you won't have LeMury's plugins... hehehe __________________ www.palcomp3.com.br/ed

Nov 2, 2003, 03:27 PM	Thread Starter #151
Lex Nahumury DH Senior Member Join Date: Jan 2003 Location: The Netherlands Posts: 1,932 Rep Power: 64	mAlo, >>I mean the ability to select in the combos of the Eplt even the asio outputs I see Well, the 'Eplg' which I released in this thread allready does that. But the design problem with having selectable ASIO channels in a 4x2 config, is that unused asio channels have to be muted to "empty" the asio buffers with zeros. This waistes ~14 extra dsp instructions, and ~2 gprs! So we loose some of the resource saving advantage, and it is not possable to load more then 1 'Eplg' because of this asio muting feature. (read my correspondance with Travelrec) So from a 'resource saving' point of view, there was no profit and I decided to make the 'Asio' plugin instead, wich offers all asio channels at the same resource "price". To conlude; Almost everything is possable. I could, for example write one single Asio channel pair plugin wich uses only 2 instr. and 2 gprs. This would be great for saving resources if you only have to record to 1 asio channel pair. But I'm afraid that releasing such plugins would confuse kX users who have little kX DSP knowledge thereby raising the nr of "how to" questions in general. Feel free to ask any questions, suggestions etc. That is what this thread is meant for. /LeMury

Nov 12, 2003, 11:29 PM	Thread Starter #158
Lex Nahumury DH Senior Member Join Date: Jan 2003 Location: The Netherlands Posts: 1,932 Rep Power: 64	>>point me to what i have to downloat for the src and the other plugins. for kx version 35. Sry, there are no downloads for 3535 anymore since they got bundled with 3536c driver release. /LeMury

Nov 13, 2003, 08:55 PM	Thread Starter #161
Lex Nahumury DH Senior Member Join Date: Jan 2003 Location: The Netherlands Posts: 1,932 Rep Power: 64	>>hmm... what if i dont want 3536? I see. Well, it turned out that a lot of (new) users don't understand how an "open" project like kX works. I.e. they simply assume that every (not included) dsp plugin should automaticly work with every driver release. If it doesn't work ppl started reporting 'driver bugs' and stuff like that. To avoid such confusion there are no more "unsupported" plugin versions pubicly available for download. However, since this obviously doesn't apply to you (as a 'kX veteran') send me an e-mail, and I'll send you the 'old' versions, or contact Daniel Drummond for a copy. Note that I will not maintain any updates nor support for these old versions. All tech-info can be found in this thread. /LeMury

Nov 13, 2003, 11:54 PM	#162
Daniel Drummond DriverHeaven Senior Member Join Date: Dec 2002 Location: Brazil Posts: 1,137 Rep Power: 0	Yes Kokoon, I can send those to you if you want. Just let me know! __________________ www.palcomp3.com.br/ed

Nov 16, 2003, 05:19 PM	#163
kokoon DriverHeaven Addict Join Date: Feb 2003 Location: slovenia Posts: 269 Rep Power: 0	well since version 3536 looks pretty stable by now, i think i'll just try it out. thanks anyway! __________________ [kokoon]

Nov 21, 2003, 08:14 AM	#164
eyagos DriverHeaven Addict Join Date: Dec 2002 Posts: 259 Rep Power: 0	Hi. There is a bug in the plugin src included in 3536c: The Line In has a lower level than epilogs one (-12dB). I have been looking, and when any AC97 source is selected, the output is multiplyed by 0.25 Regards. Last edited by eyagos; Nov 21, 2003 at 08:22 AM.

Nov 21, 2003, 08:50 AM	Thread Starter #165
Lex Nahumury DH Senior Member Join Date: Jan 2003 Location: The Netherlands Posts: 1,932 Rep Power: 64	>>There is a bug in the plugin src included in 3536c: The Line In has a lower level than >> epilogs one (-12dB). I have been looking, and when any AC97 source is selected, the output is >>multiplyed by 0.25 First; You probably mean Prolog instead of Epilog, since Epilog does not do AC97 input at all. Second; Src's AC97 inputs uses NO digital attenuation/amplification like the new Prolog does!!! This is intentional to give better SNR. (the 0x20000000 mult is standard) Since 3535 now you can get ~-75dB noise floor instead of ~-63dB. (please read recent threads about ac97 input levels, SNR etc) I hope it's clear now. (nevertheless I will check it) Regards, /LeMury Last edited by LeMury; Nov 21, 2003 at 09:14 AM.

Nov 21, 2003, 12:30 PM	#166
eyagos DriverHeaven Addict Join Date: Dec 2002 Posts: 259 Rep Power: 0	First: I not only have readed the recent threads, but I have checked too that multiplying by 4 were incorrect. I normally feed into the Line In the signal from a TV card and I got a maximum level of 0dB. But feedind in it the HI-FI, wich can provide more voltage, I can get +12dB, so the plugin is correct. Now I ask myself why in 3536c (the version I'm ussing now) prolog all AC97 inputs are multiplyed by 0x2000000, except Line In wich is multiplyed by 0x7FFFFFFF. And other thing, although I understand you, you are not talking well when you say that SNR is increased by multiplying by 0.25, because you are reducing noise and signal at the same amount. What you must say is that you are increasing the signal (and noise) range playable in the speakers. And second: Of course I wanted to say 'epilog' when I sayd 'epilog'. Sorry, prolog Regards. Last edited by eyagos; Nov 21, 2003 at 12:39 PM.

Nov 21, 2003, 01:30 PM	Thread Starter #167
Lex Nahumury DH Senior Member Join Date: Jan 2003 Location: The Netherlands Posts: 1,932 Rep Power: 64	>>you are not talking well when you say that SNR is increased by multiplying by 0.25, Sorry Eyagos, but that is not what I said. The SNR of 'kX's ac97's ADC to DSP signal path' has increased since 3535 because there was an 'invalid' (digital) amplification after ADC stage. Eugene/Max please FIXME here..! Sorry Eyagos, I'm just getting tired of the subject since I already posted many posts with extensive details as did others. Regards, /LeMury

Nov 21, 2003, 03:53 PM	#168
eyagos DriverHeaven Addict Join Date: Dec 2002 Posts: 259 Rep Power: 0	Don't bother what you have concluded in other posts with other people, but SNR has not been increased at all: My TV card with 'old' drivers -> Floor = -60dB = 0.001 // Maximum = 0dB = 1 // SNR = 60dB My TV card with 'new' drivers -> Floor = -72dB = 0.00025 // Maximum = -12dB = 0.25 // SNR = 60dB If we are talking about 'the same SNR' ...

Nov 21, 2003, 04:50 PM	Thread Starter #169
Lex Nahumury DH Senior Member Join Date: Jan 2003 Location: The Netherlands Posts: 1,932 Rep Power: 64	>>My TV card with 'old' drivers -> Floor = -60dB = 0.001 // Maximum = 0dB = 1 // SNR = 60dB >>My TV card with 'new' drivers -> Floor = -72dB = 0.00025 // Maximum = -12dB = 0.25 // SNR = 60dB Yep, the above is correct. Only,..The difference now is that you can Increase the analog input signal to get to the same 0dB, but without digitally amplifying after A/D conversion. I.E. If you would take a pre-amp and amplify your TV-card's analog signal so that it uses ADC's full range you will get 0dB again..! But this time without digitaly amplifying ac97's noise floor..! So the SNR of AC97 input in kX has been improved. In your case, if you leave the analog TV-card signal the same, there is NO SNR improvement because you will (again) digitaly amplify just like it was before 3535. In (yet) other words; We do not have to digitaly amplify ac97's noise floor if we make sure our analog input signal is high enough to get 0dB. /LeMury

Nov 21, 2003, 05:55 PM	#170
eyagos DriverHeaven Addict Join Date: Dec 2002 Posts: 259 Rep Power: 0	yes yes yes. We say the same: Now we can achieve a -72dB floor noise. Suposing that these are the minimum that can be achieved with an Audigy2 (mine), and knowing that 'SigmaTel STAC9723' (mine) gives -90dB SNR, we can say that the Audigy introduces 18dB of aditional noise. Then, let's say that Audigy2's AC97 inputs (mine) have a -72dB SNR. Okey, but drivers can't increase/decrease this SNR!!!!!!! It's only that yesterday these SNR dBs where distributed between [-60dB, +12dB] (and then loosing that 12dB becuse of clipping) , and today are distributed between [-72dB, 0dB] (what is more correct). But the noise HEARD inside a 'clean' signal is the same today than yesterday or, in other words, the heard sound has the same quality, or the same SNR (think in a -20dB clean signal). Do you understand what I mean now?.

Nov 21, 2003, 07:34 PM	Thread Starter #171
Lex Nahumury DH Senior Member Join Date: Jan 2003 Location: The Netherlands Posts: 1,932 Rep Power: 64	>>Then, let's say that Audigy2's AC97 inputs (mine) have a -72dB SNR. >> Okey, but drivers can't increase/decrease this SNR!!!!!!! No, of course not. This is defined in hardware. >>It's only that yesterday these SNR dBs where distributed between [-60dB, +12dB] >> (and then loosing that 12dB becuse of clipping) >> and today are distributed between [-72dB, 0dB] (what is more correct). "Today" the analog Input signal, relative to ac97's noise floor, can be set ~10db higher without clipping. To me that translates as an SNR improvement. (or SNR correction, if you wish) I have no other word for it. But realy Eyagos,.. we probably mean/say the same but only interprete it differently. I understood your point but if you don't mind, let's drop this discussion. All I'm interested here in this thread is to make plugins that are within kX's I/O specs. Regards, /LeMury

Nov 21, 2003, 09:03 PM	#172
Max M. h/h member-shmember Join Date: Dec 2002 Location: Evil Empire Posts: 2,639 Rep Power: 69	>But realy Eyagos,.. we probably mean/say the same but only interprete it differently. yep, it seems like that. Although it would be more correct to say that "older versions just forced user to choose not correct settings with worse SNR" (and user can get the same signal path as it has now by setting "ac97" fader to 25%)... E.g. newer driver versions do not improve SNR itself but "force the user to deal with initially better setting which gives better SNR"... something like that... >knowing that 'SigmaTel STAC9723' (mine) gives -90dB SNR, we can say that the Audigy introduces 18dB of aditional noise. Well, i'd said that these values written in codec specification are more like "abstract values", like "the best expected SNR in ideal enviroment" (when chip is not in machine and when it is not connected to anywhere at all)... "that the Audigy introduces 18dB of aditional noise" - hmm, no, i would not say that... __________________