[Csnd] audio software idiosyncrasies

Date

2010-04-12 16:36

From

kind beta

Subject

[Csnd] audio software idiosyncrasies

hello.

i have an issue but it seems difficult to me to make a precise question.
so i'll try to describe:

every audiosoftware i use sounds idiosyncratic. (mainly i use msp, reaktor for a long time now and csound since last year).

certainly there is a huge influence of the software, of its capabilities and its gui to the working/composing process. but it seems to me like there are drastic differences in the way the software handles audio signals. the most simple example i can think of is distortion introduced buy amplitude override. just take a normalised audiofile, and plyback it through a software by multiplying its amplitude by a number >1. this will produce a different sound from software to software. i guess everyone experienced this.

this specific sound coloring that it introduces is something i take into account for the decision which software to use to achieve a imagined soundquality. now i would like to know some theory behind this. to have words for what i hear...

i guess this is due things like word depths and the way the software accomplish truncation, filtering... but is there any specific knowlege, experience, writings, comparisons or discussins on this subject?

best regards
friedrich

__________________________________________________
Do You Yahoo!?
Sie sind Spam leid? Yahoo! Mail verfügt über einen herausragenden Schutz gegen Massenmails.
http://mail.yahoo.com

Date

2010-04-12 19:26

From

Michael Gogins

Subject

[Csnd] Re: audio software idiosyncrasies

This is not a simple issue, and and an accurate answer is not possible without knowing technical and perhaps private details of each system.

However, there are some factors that are common to all systems.

If the signal is within -3 dB of full scale, even a 16 bit signal is going to sound just fine, that is not quite as good as human hearing but not far below it, and 24 bits or float samples can easily exceed the resolution of human hearing. At that precision, the electronics are more limiting than the software.

The behavior of the signal if there is clipping will depend upon the software. Csound will do nothing unless you tell it to. I imagine the same is true of Max/MSP. Other commercial software may use limiting and compression to mitigate the bad sound of clipping.

If there is clipping without limiting, then the sound may possibly depend upon the audio driver software, and will certainly depend upon the electronics and the speakers. With good electronics and speakers and no limiting, the clipping will sound really terrible, like ripping cloth or torn metal. If there is limiting, there may be some of that but the loud parts will mostly just be kind of muffled.

Other differences that you may hear may have nothing to do with clipping, but may depend on the synthesis algorithms or other details of the signal processing software. If you do not use bandlimited oscillators there may be noise from aliasing. Analog filters do not sound like digital filters, and different digital filters can sound quite different. And so on.

I can assure that with Csound and probably with Max/MSP, you should be able to achieve results on a part with the very best analog studios in the world, but only if you take care and understand what you are doing: using bandlimited oscillators, using the right kind of filters, taking special care with reverb, and so on. With this kind of software, there is no protection built in and it is more like a toolkit full of sharp edges that can produce exactly what you need -- but if you do not know what you are doing, you will cut yourself by leaving out some essential step.

I recommend Bob Katz' book "Mastering Audio" for a general introduction to sound quality. He explains the issues that I have mentioned in a very clear and detailed way.

I also have written an introduction to audio quality in Csound that can be found at the end of my Csound tutorial in the Windows installers for Csound.

Regards,
Mike

On Mon, Apr 12, 2010 at 11:36 AM, kind beta <kindbeta@yahoo.de> wrote:

hello.

i have an issue but it seems difficult to me to make a precise question.
so i'll try to describe:

every audiosoftware i use sounds idiosyncratic. (mainly i use msp, reaktor for a long time now and csound since last year).

certainly there is a huge influence of the software, of its capabilities and its gui to the working/composing process. but it seems to me like there are drastic differences in the way the software handles audio signals. the most simple example i can think of is distortion introduced buy amplitude override. just take a normalised audiofile, and plyback it through a software by multiplying its amplitude by a number >1. this will produce a different sound from software to software. i guess everyone experienced this.

this specific sound coloring that it introduces is something i take into account for the decision which software to use to achieve a imagined soundquality. now i would like to know some theory behind this. to have words for what i hear...

i guess this is due things like word depths and the way the software accomplish truncation, filtering... but is there any specific knowlege, experience, writings, comparisons or discussins on this subject?

best regards
friedrich

__________________________________________________
Do You Yahoo!?
Sie sind Spam leid? Yahoo! Mail verfügt über einen herausragenden Schutz gegen Massenmails.
http://mail.yahoo.com

--
Michael Gogins
Irreducible Productions
http://www.michael-gogins.com
Michael dot Gogins at gmail dot com

Date

2010-04-12 19:38

From

Peiman Khosravi

Subject

[Csnd] Re: Re: audio software idiosyncrasies

Also note that the double precision build of csound offers much higher more precision (in the DSP calculation not to be confused with bit depth) than msp and you can certainly hear the difference in some cases (particularly in the realm of FFT analysis/transformation/re-synthesis). One reason why I always use csound as my DSP engine and Max for interface building. (of course it depends on the task).

Best,

On 12 Apr 2010, at 19:26, Michael Gogins wrote:

This is not a simple issue, and and an accurate answer is not possible without knowing technical and perhaps private details of each system.

However, there are some factors that are common to all systems.

If the signal is within -3 dB of full scale, even a 16 bit signal is going to sound just fine, that is not quite as good as human hearing but not far below it, and 24 bits or float samples can easily exceed the resolution of human hearing. At that precision, the electronics are more limiting than the software.

The behavior of the signal if there is clipping will depend upon the software. Csound will do nothing unless you tell it to. I imagine the same is true of Max/MSP. Other commercial software may use limiting and compression to mitigate the bad sound of clipping.

If there is clipping without limiting, then the sound may possibly depend upon the audio driver software, and will certainly depend upon the electronics and the speakers. With good electronics and speakers and no limiting, the clipping will sound really terrible, like ripping cloth or torn metal. If there is limiting, there may be some of that but the loud parts will mostly just be kind of muffled.

Other differences that you may hear may have nothing to do with clipping, but may depend on the synthesis algorithms or other details of the signal processing software. If you do not use bandlimited oscillators there may be noise from aliasing. Analog filters do not sound like digital filters, and different digital filters can sound quite different. And so on.

I can assure that with Csound and probably with Max/MSP, you should be able to achieve results on a part with the very best analog studios in the world, but only if you take care and understand what you are doing: using bandlimited oscillators, using the right kind of filters, taking special care with reverb, and so on. With this kind of software, there is no protection built in and it is more like a toolkit full of sharp edges that can produce exactly what you need -- but if you do not know what you are doing, you will cut yourself by leaving out some essential step.

I recommend Bob Katz' book "Mastering Audio" for a general introduction to sound quality. He explains the issues that I have mentioned in a very clear and detailed way.

I also have written an introduction to audio quality in Csound that can be found at the end of my Csound tutorial in the Windows installers for Csound.

Regards,
Mike

On Mon, Apr 12, 2010 at 11:36 AM, kind beta <kindbeta@yahoo.de> wrote:

hello.

i have an issue but it seems difficult to me to make a precise question.
so i'll try to describe:

every audiosoftware i use sounds idiosyncratic. (mainly i use msp, reaktor for a long time now and csound since last year).

certainly there is a huge influence of the software, of its capabilities and its gui to the working/composing process. but it seems to me like there are drastic differences in the way the software handles audio signals. the most simple example i can think of is distortion introduced buy amplitude override. just take a normalised audiofile, and plyback it through a software by multiplying its amplitude by a number >1. this will produce a different sound from software to software. i guess everyone experienced this.

this specific sound coloring that it introduces is something i take into account for the decision which software to use to achieve a imagined soundquality. now i would like to know some theory behind this. to have words for what i hear...

i guess this is due things like word depths and the way the software accomplish truncation, filtering... but is there any specific knowlege, experience, writings, comparisons or discussins on this subject?

best regards
friedrich

__________________________________________________
Do You Yahoo!?
Sie sind Spam leid? Yahoo! Mail verfügt über einen herausragenden Schutz gegen Massenmails.
http://mail.yahoo.com

--
Michael Gogins
Irreducible Productions
http://www.michael-gogins.com
Michael dot Gogins at gmail dot com

Date

2010-04-12 19:48

From

Anthony Palomba

Subject

[Csnd] Re: Re: Re: audio software idiosyncrasies

I use Reaktor quite a bit, and I know for a fact that they are doing a
lot under the hood to make things sound better. When you add a
macro to your ensemble you may not be aware of all it is doing, so
when you try a similar thing in say Max, you get a different sounding
output. It is pretty annoying to be honest with you.

But you can actually learn a lot from trying rebuild Reaktor macros in Max
or csound. It takes some investigating and experimenting. After a while
you start to get create an awareness of the different techniques used to make things
sound better. If you get the Katz book and start reading, you'll see that
making things sound better is an art all it's own.

-ap

On Mon, Apr 12, 2010 at 1:38 PM, Peiman Khosravi <peimankhosravi@gmail.com> wrote:

Also note that the double precision build of csound offers much higher more precision (in the DSP calculation not to be confused with bit depth) than msp and you can certainly hear the difference in some cases (particularly in the realm of FFT analysis/transformation/re-synthesis). One reason why I always use csound as my DSP engine and Max for interface building. (of course it depends on the task).

Best,

P

On 12 Apr 2010, at 19:26, Michael Gogins wrote:

This is not a simple issue, and and an accurate answer is not possible without knowing technical and perhaps private details of each system.

However, there are some factors that are common to all systems.

If the signal is within -3 dB of full scale, even a 16 bit signal is going to sound just fine, that is not quite as good as human hearing but not far below it, and 24 bits or float samples can easily exceed the resolution of human hearing. At that precision, the electronics are more limiting than the software.

The behavior of the signal if there is clipping will depend upon the software. Csound will do nothing unless you tell it to. I imagine the same is true of Max/MSP. Other commercial software may use limiting and compression to mitigate the bad sound of clipping.

If there is clipping without limiting, then the sound may possibly depend upon the audio driver software, and will certainly depend upon the electronics and the speakers. With good electronics and speakers and no limiting, the clipping will sound really terrible, like ripping cloth or torn metal. If there is limiting, there may be some of that but the loud parts will mostly just be kind of muffled.

Other differences that you may hear may have nothing to do with clipping, but may depend on the synthesis algorithms or other details of the signal processing software. If you do not use bandlimited oscillators there may be noise from aliasing. Analog filters do not sound like digital filters, and different digital filters can sound quite different. And so on.

I can assure that with Csound and probably with Max/MSP, you should be able to achieve results on a part with the very best analog studios in the world, but only if you take care and understand what you are doing: using bandlimited oscillators, using the right kind of filters, taking special care with reverb, and so on. With this kind of software, there is no protection built in and it is more like a toolkit full of sharp edges that can produce exactly what you need -- but if you do not know what you are doing, you will cut yourself by leaving out some essential step.

I recommend Bob Katz' book "Mastering Audio" for a general introduction to sound quality. He explains the issues that I have mentioned in a very clear and detailed way.

I also have written an introduction to audio quality in Csound that can be found at the end of my Csound tutorial in the Windows installers for Csound.

Regards,
Mike

On Mon, Apr 12, 2010 at 11:36 AM, kind beta <kindbeta@yahoo.de> wrote:

hello.

i have an issue but it seems difficult to me to make a precise question.
so i'll try to describe:

every audiosoftware i use sounds idiosyncratic. (mainly i use msp, reaktor for a long time now and csound since last year).

certainly there is a huge influence of the software, of its capabilities and its gui to the working/composing process. but it seems to me like there are drastic differences in the way the software handles audio signals. the most simple example i can think of is distortion introduced buy amplitude override. just take a normalised audiofile, and plyback it through a software by multiplying its amplitude by a number >1. this will produce a different sound from software to software. i guess everyone experienced this.

this specific sound coloring that it introduces is something i take into account for the decision which software to use to achieve a imagined soundquality. now i would like to know some theory behind this. to have words for what i hear...

i guess this is due things like word depths and the way the software accomplish truncation, filtering... but is there any specific knowlege, experience, writings, comparisons or discussins on this subject?

best regards
friedrich

__________________________________________________
Do You Yahoo!?
Sie sind Spam leid? Yahoo! Mail verfügt über einen herausragenden Schutz gegen Massenmails.
http://mail.yahoo.com

--
Michael Gogins
Irreducible Productions
http://www.michael-gogins.com
Michael dot Gogins at gmail dot com

Date

2010-04-12 21:51

From

kind beta

Subject

[Csnd] AW: Re: Re: Re: audio software idiosyncrasies

thanks for the responses.

@mike: it's certainly difficult or unpossible to give an accurate answer since i rather failed at even drafting out a question. my post was rather a kind of an appeal for exchange of experiences i guess.
starting with csound i read your tutorial and helped a lot thanks. i'm not sure if it was in your tutorial or in the csound book that the high word depths of csoud calculations were mentioned... the interessting thing for me was that the specific sound resulting from this more accurate dsp calculations were described as hyper realistic (or something like that). the use of this nearly illustrative term together with the technical explanation, was exactly what helped me to understand and conect that knowlege to my thoughts about composition and sound shaping (since i'm much more software user than developer)

@peiman: the csound/max thing has become one of my prefered configurations during the last months. but for some fft tasks i sometimes prefere some msp patches especially due to their implied roughness, grain or the high glass/water like sounding artefacts.

@ap: yes, all the ni software i listened to seem to have a very idiosyncratic sound due to some evidently unvisible shaping. and using reaktor from the more developer perspective one could be annoyed by the loss of controll/freedom implied by this saturation/filtering routines... as a more user than developer i sometimes tend to look at this like using a specific instrument. you know, like a guitar player deciding to use a les paul model when he preferes a more warm and grounded sound, and a tellecaster for more bright and shiny colors. and he knows this sound is due to the material, guitarbody and the pickups and so on. this is a problematic comparison of course, since computer musicans face their specific problemms which can not be fully undersood through such analogies.

@all: as michael said the mastering audio book is a good technical general introduction. but isn't it a huge effort to read it due to its imperative statments about good and bad sound? for me its narrow esthetic focus is kind of paralysing.

thank you
f.

--- Anthony Palomba <apalomba@austin.rr.com> schrieb am Mo, 12.4.2010:

Von: Anthony Palomba <apalomba@austin.rr.com>
Betreff: [Csnd] Re: Re: Re: audio software idiosyncrasies
An: csound@lists.bath.ac.uk
Datum: Montag, 12. April, 2010 20:48 Uhr

I use Reaktor quite a bit, and I know for a fact that they are doing a
lot under the hood to make things sound better. When you add a
macro to your ensemble you may not be aware of all it is doing, so
when you try a similar thing in say Max, you get a different sounding
output. It is pretty annoying to be honest with you.

But you can actually learn a lot from trying rebuild Reaktor macros in Max
or csound. It takes some investigating and experimenting. After a while
you start to get create an awareness of the different techniques used to make things
sound better. If you get the Katz book and start reading, you'll see that
making things sound better is an art all it's own.

-ap

On Mon, Apr 12, 2010 at 1:38 PM, Peiman Khosravi <peimankhosravi@gmail.com> wrote:

Also note that the double precision build of csound offers much higher more precision (in the DSP calculation not to be confused with bit depth) than msp and you can certainly hear the difference in some cases (particularly in the realm of FFT analysis/transformation/re-synthesis). One reason why I always use csound as my DSP engine and Max for interface building. (of course it depends on the task).

Best,

P

On 12 Apr 2010, at 19:26, Michael Gogins wrote:

This is not a simple issue, and and an accurate answer is not possible without knowing technical and perhaps private details of each system.

However, there are some factors that are common to all systems.

If the signal is within -3 dB of full scale, even a 16 bit signal is going to sound just fine, that is not quite as good as human hearing but not far below it, and 24 bits or float samples can easily exceed the resolution of human hearing. At that precision, the electronics are more limiting than the software.

The behavior of the signal if there is clipping will depend upon the software. Csound will do nothing unless you tell it to. I imagine the same is true of Max/MSP. Other commercial software may use limiting and compression to mitigate the bad sound of clipping.

If there is clipping without limiting, then the sound may possibly depend upon the audio driver software, and will certainly depend upon the electronics and the speakers. With good electronics and speakers and no limiting, the clipping will sound really terrible, like ripping cloth or torn metal. If there is limiting, there may be some of that but the loud parts will mostly just be kind of muffled.

Other differences that you may hear may have nothing to do with clipping, but may depend on the synthesis algorithms or other details of the signal processing software. If you do not use bandlimited oscillators there may be noise from aliasing. Analog filters do not sound like digital filters, and different digital filters can sound quite different. And so on.

I can assure that with Csound and probably with Max/MSP, you should be able to achieve results on a part with the very best analog studios in the world, but only if you take care and understand what you are doing: using bandlimited oscillators, using the right kind of filters, taking special care with reverb, and so on. With this kind of software, there is no protection built in and it is more like a toolkit full of sharp edges that can produce exactly what you need -- but if you do not know what you are doing, you will cut yourself by leaving out some essential step.

I recommend Bob Katz' book "Mastering Audio" for a general introduction to sound quality. He explains the issues that I have mentioned in a very clear and detailed way.

I also have written an introduction to audio quality in Csound that can be found at the end of my Csound tutorial in the Windows installers for Csound.

Regards,
Mike

On Mon, Apr 12, 2010 at 11:36 AM, kind beta <kindbeta@yahoo.de> wrote:

hello.

i have an issue but it seems difficult to me to make a precise question.
so i'll try to describe:

every audiosoftware i use sounds idiosyncratic. (mainly i use msp, reaktor for a long time now and csound since last year).

certainly there is a huge influence of the software, of its capabilities and its gui to the working/composing process. but it seems to me like there are drastic differences in the way the software handles audio signals. the most simple example i can think of is distortion introduced buy amplitude override. just take a normalised audiofile, and plyback it through a software by multiplying its amplitude by a number >1. this will produce a different sound from software to software. i guess everyone experienced this.

this specific sound coloring that it introduces is something i take into account for the decision which software to use to achieve a imagined soundquality. now i would like to know some theory behind this. to have words for what i hear...

i guess this is due things like word depths and the way the software accomplish truncation, filtering... but is there any specific knowlege, experience, writings, comparisons or discussins on this subject?

best regards
friedrich

__________________________________________________
Do You Yahoo!?
Sie sind Spam leid? Yahoo! Mail verfügt über einen herausragenden Schutz gegen Massenmails.
http://mail.yahoo.com

--
Michael Gogins
Irreducible Productions
http://www.michael-gogins.com
Michael dot Gogins at gmail dot com

__________________________________________________
Do You Yahoo!?
Sie sind Spam leid? Yahoo! Mail verfügt über einen herausragenden Schutz gegen Massenmails.
http://mail.yahoo.com

Date	2010-05-19 22:14
From	Brian Wong
Subject	[Csnd] Tritium Tea
	Three tempered tritave tunings take turns. http://bwong.ca/template1.php?sub=11 Brian Wong 30 days of prizes to be won with Hotmail. Enter here