Super Wav? Super Flac?

There are high-definition CD and DVD audio disk standards now
competing for market share with ultra-high-fidelity disks (which you
can't copy, so if one gets munged you are out of luck.) I'm wondering
if anyone has given any thought to higher-fidelity digital file standards?

Of course, this is pretty futuristic for microtonalists given the
audio quality we generally deal with, but I'm wondering anyway.

On 12/18/06, Gene Ward Smith <genewardsmith@...> wrote:
> There are high-definition CD and DVD audio disk standards now
> competing for market share with ultra-high-fidelity disks (which you
> can't copy, so if one gets munged you are out of luck.) I'm wondering
> if anyone has given any thought to higher-fidelity digital file standards?
>
> Of course, this is pretty futuristic for microtonalists given the
> audio quality we generally deal with, but I'm wondering anyway.

I think you're assuming RIFF WAVE and FLAC have restrictions they
don't really have. RIFF WAVE files don't have to be 16 bits and 44100
samples per second. The number of bits per sample can be any 16-bit
number (so the maximum is 65535 bits per sample) and the sampling rate
can be any 32-bit number (so the maximum sampling rate is 4.29
gigahertz). FLAC has similar flexibility.

Keenan

>There are high-definition CD and DVD audio disk standards now
>competing for market share with ultra-high-fidelity disks (which you
>can't copy, so if one gets munged you are out of luck.) I'm wondering
>if anyone has given any thought to higher-fidelity digital file standards?

I routinely recorded to 24-bit/48Khz. and saved the files as
FLACs. That is, before I left anything resembling a music-related
career to work in Silicone Valley (as I like to call it). Now
I don't do #&$*.

24/48 is I think the limit of reasonable fidelity. The sensible
thing to improve at that point is the age-old notion of "stereo".
In particular, the recording should have metadata that tells how
it was recorded, and the playback system should attempt to do the
best it can with that. Here's one attempt at this

http://lumma.org/stuff/AudioReproduction.xls

-Carl

Gene Ward Smith escreveu:
> There are high-definition CD and DVD audio disk standards now
> competing for market share with ultra-high-fidelity disks

Any references?

> (which you
> can't copy,

Why not?

--
'-------------------------------------------------------------------.
Hudson Lacerda http://br.geocities.com/hfmlacerda/

*THE WAR IN IRAQ COSTS: http://costofwar.com/

microabc -- free software for microtonal music
http://br.geocities.com/hfmlacerda/abc/microabc-about.html

*N�O DEIXE SEU VOTO SUMIR! http://www.votoseguro.org/
*Ap�ie o Manifesto: http://www.votoseguro.com/alertaprofessores/
.-------------------------------------------------------------------'
--

_______________________________________________________ Yahoo! Mail - Sempre a melhor op��o para voc�! Experimente j� e veja as novidades. http://br.yahoo.com/mailbeta/tudonovo/

Hi Hudson,

>> (which you can't copy,
>
>Why not?

In the case of SACD... sigma-delta modulation is used, which is
not supported by any PC hardware or sofware that I know of.

Despite a fair amount of titles available for SACD, the format
has been somewhat stillborn. Ditto DVD-Audio. That's because
even the sub-CD quality of most mp3 files is plenty good enough
for most people.

-Carl

Carl Lumma escreveu:
> Hi Hudson,
> > >>>(which you can't copy,
>>
>>Why not?
> > > In the case of SACD... sigma-delta modulation is used, which is
> not supported by any PC hardware or sofware that I know of.

So it cannot be read in PCs, but only in specialised hardware...

We have not recorders, we cannot record music is such a media, thus it is completely useless! :-)

And, since I am opposed to any form of DRM, my interest on such discs is null so far.

Thanks for the info, Carl!

Hudson

> > Despite a fair amount of titles available for SACD, the format
> has been somewhat stillborn. Ditto DVD-Audio. That's because
> even the sub-CD quality of most mp3 files is plenty good enough
> for most people.
> > -Carl
> > > > > Yahoo! Groups Links
> > > > -- '-------------------------------------------------------------------.
Hudson Lacerda http://br.geocities.com/hfmlacerda/

*THE WAR IN IRAQ COSTS: http://costofwar.com/

microabc -- free software for microtonal music
http://br.geocities.com/hfmlacerda/abc/microabc-about.html

*N�O DEIXE SEU VOTO SUMIR! http://www.votoseguro.org/
*Ap�ie o Manifesto: http://www.votoseguro.com/alertaprofessores/
.-------------------------------------------------------------------'
--

_______________________________________________________ Yahoo! Mail - Sempre a melhor op��o para voc�! Experimente j� e veja as novidades. http://br.yahoo.com/mailbeta/tudonovo/

>> In the case of SACD... sigma-delta modulation is used, which is
>> not supported by any PC hardware or sofware that I know of.
>
>We have not recorders, we cannot record music is such a media, thus it
>is completely useless! :-)
>
>And, since I am opposed to any form of DRM, my interest on such discs
>is null so far.

I am also completely opposed to DRM. Delta-sigma (I got it backward
above) sampling, though, isn't a form of DRM, just a different form of
digital sampling that some believe to be superior to PCM.

However, despite that no consumer hardware exists for decoding such
a format, Sony still insists on encrypting the digital output of
their SACD players!

-Carl

--- In MakeMicroMusic@yahoogroups.com, Carl Lumma <ekin@...> wrote:
>
> >> In the case of SACD... sigma-delta modulation is used, which is
> >> not supported by any PC hardware or sofware that I know of.
> >
> >We have not recorders, we cannot record music is such a media, thus it
> >is completely useless! :-)
> >
> >And, since I am opposed to any form of DRM, my interest on such discs
> >is null so far.
>
> I am also completely opposed to DRM. Delta-sigma (I got it backward
> above) sampling, though, isn't a form of DRM, just a different form of
> digital sampling that some believe to be superior to PCM.
>
> However, despite that no consumer hardware exists for decoding such
> a format, Sony still insists on encrypting the digital output of
> their SACD players!

SONY are assholes. I'm still pissed that I can't simply drag and drop
my wav files that I recorded in my own environment from my Sony
mini-disc recorder. You have to have proprietary software, and even
then, they only allow you to do it once--I gues they are afraid of
people copying all the premade Sony mini-discs of Sony artists out
there (irony intended).

What is their problem? One should be able to use the mini-disc as a
USB storage device and do a harmless transfer. Instead, to get my
samples in my computer, I have to go analog, real-time into the sound
card, which sucks.

-A.

Boys,

Super OT if you ask me.

Cheers,
Jon

--- In MakeMicroMusic@yahoogroups.com, Carl Lumma <ekin@...> wrote:

> 24/48 is I think the limit of reasonable fidelity. The sensible
> thing to improve at that point is the age-old notion of "stereo".
> In particular, the recording should have metadata that tells how
> it was recorded, and the playback system should attempt to do the
> best it can with that. Here's one attempt at this
>
> http://lumma.org/stuff/AudioReproduction.xls

The point of my question is that I thought the new ultra-hi-fi was
attempting to do something more than improve on what 44100 stereo
gives you in more ways than upping the bit rate. Is that wrong?

--- In MakeMicroMusic@yahoogroups.com, Carl Lumma <ekin@...> wrote:

> However, despite that no consumer hardware exists for decoding such
> a format, Sony still insists on encrypting the digital output of
> their SACD players!

Then they fall down the analog hole anyway.

--- In MakeMicroMusic@yahoogroups.com, Jon Szanto <jszanto@...> wrote:
>
> Boys,
>
> Super OT if you ask me.

I think formats of the music we produce is entirely on topic.

>> 24/48 is I think the limit of reasonable fidelity. The sensible
>> thing to improve at that point is the age-old notion of "stereo".
>> In particular, the recording should have metadata that tells how
>> it was recorded, and the playback system should attempt to do the
>> best it can with that. Here's one attempt at this
>>
>> http://lumma.org/stuff/AudioReproduction.xls
>
>The point of my question is that I thought the new ultra-hi-fi was
>attempting to do something more than improve on what 44100 stereo
>gives you in more ways than upping the bit rate. Is that wrong?

Is it wrong to do it? No. Is it the logical, obvious, or best
way to do it? No.

-Carl

>SONY are assholes.

That pretty-much sums it up.

-Carl

>> However, despite that no consumer hardware exists for decoding such
>> a format, Sony still insists on encrypting the digital output of
>> their SACD players!
>
>Then they fall down the analog hole anyway.

The whole point of SACD is quality > CD audio. What would be the
point of an analog copy of SACD audio??

-Carl

--- In MakeMicroMusic@yahoogroups.com, Carl Lumma <ekin@...> wrote:

> >The point of my question is that I thought the new ultra-hi-fi was
> >attempting to do something more than improve on what 44100 stereo
> >gives you in more ways than upping the bit rate. Is that wrong?
>
> Is it wrong to do it? No. Is it the logical, obvious, or best
> way to do it? No.

What I was asking was whether I was wrong in thinking more than an
increased bitrate was involved, and what might be done in the future
involving using these capabilities for music we might make.

--- In MakeMicroMusic@yahoogroups.com, Carl Lumma <ekin@...> wrote:
>
> >> However, despite that no consumer hardware exists for decoding such
> >> a format, Sony still insists on encrypting the digital output of
> >> their SACD players!
> >
> >Then they fall down the analog hole anyway.
>
> The whole point of SACD is quality > CD audio. What would be the
> point of an analog copy of SACD audio??

If you have a very high fidelity signal, I presume you could save it
as 24-bit flac or whatever.

At 02:22 PM 12/19/2006, you wrote:
>--- In MakeMicroMusic@yahoogroups.com, Carl Lumma <ekin@...> wrote:
>>
>> >> However, despite that no consumer hardware exists for decoding such
>> >> a format, Sony still insists on encrypting the digital output of
>> >> their SACD players!
>> >
>> >Then they fall down the analog hole anyway.
>>
>> The whole point of SACD is quality > CD audio. What would be the
>> point of an analog copy of SACD audio??
>
>If you have a very high fidelity signal, I presume you could save it
>as 24-bit flac or whatever.

That won't be as good as the DSD signal on the SACD, for
three reasons:

1. 24-bit PCM isn't as good as DSD, at least if you believe Sony.
2. You're getting the D/A and then A/D conversion errors.
3. You're getting whatever losses are associated with your
analog cable.

-Carl

>> >The point of my question is that I thought the new ultra-hi-fi was
>> >attempting to do something more than improve on what 44100 stereo
>> >gives you in more ways than upping the bit rate. Is that wrong?
>>
>> Is it wrong to do it? No. Is it the logical, obvious, or best
>> way to do it? No.
>
>What I was asking was whether I was wrong in thinking more than an
>increased bitrate was involved, and what might be done in the future
>involving using these capabilities for music we might make.

There's debate about it. In some circles, the superiority of
96 KHz. sampling is taken for granted. In others, it isn't.
24-bit samples are definitely better for mastering, so that
rounding errors during editing don't make it up into the 16
most significant bits.

I can send you some papers off-list if you're interested in
the finer points of this topic.

-Carl

G,

{you wrote...}
>> Super OT if you ask me.
>
>I think formats of the music we produce is entirely on topic.

Well, it was a bit tongue-in-cheek, although when you get to informative posts like "Sony are assholes" you have to wonder. Ah, what the hell...

Cheers,
Jon

--- In MakeMicroMusic@yahoogroups.com, Jon Szanto <jszanto@...> wrote:
>
> G,
>
> {you wrote...}
Jon wrote:
> >> Super OT if you ask me.
> >
Gene wrote:
> >I think formats of the music we produce is entirely on topic.
>
Jon:
> Well, it was a bit tongue-in-cheek, although when you get to
informative posts like "Sony are assholes" you have to wonder. Ah,
what the hell...

Aaron:
Well, maybe you don't like the a-hole part, but it's the truth as I
see it. They are the reigning schmucks of the DRM movement.

In that sense, if any one on this list is thinking of buying Sony
equipment to "Make Micro Music" with, this musician will tell them
that if they actually want to optimally *use* their equipment, (as in,
have the right to upload the sample they just made into their
sequencer or other such useful musical activity), they should stay
away from Sony products.

Is that better, Jon? ;)

-A.

Carl Lumma wrote:

> There's debate about it. In some circles, the superiority of
> 96 KHz. sampling is taken for granted. In others, it isn't.
> 24-bit samples are definitely better for mastering, so that
> rounding errors during editing don't make it up into the 16
> most significant bits.

96kHz sampling has a valid application for recording analog audio to be used at 48kHz. It means you can apply a digital filter and down-sample to get the true 48kHz signal.

Beats can get lost with low sample rates, so this may be important for some uses of alternative tunings.

Csound uses floating point numbers for audio, which is even better for mastering because you don't have to worry about maximizing the levels. You can record audio files in floating point format. There are optional 64-bit builds as well. If you can hear the difference over 32 bits (some claim to) it's probably because of bugs in the system. But it doesn't slow things down much.

> I can send you some papers off-list if you're interested in
> the finer points of this topic.

I swear I could hear differences in my Kyma distortion up to 64 kHz. I didn't try down-sampling it. There may have been a placebo effect.

Ambisonics is another thing to look at. I haven't, seriously, because I don't have the speakers.

Graham

At 08:41 PM 12/19/2006, you wrote:
>Carl Lumma wrote:
>
>> There's debate about it. In some circles, the superiority of
>> 96 KHz. sampling is taken for granted. In others, it isn't.
>> 24-bit samples are definitely better for mastering, so that
>> rounding errors during editing don't make it up into the 16
>> most significant bits.
>
>96kHz sampling has a valid application for recording analog audio to be
>used at 48kHz. It means you can apply a digital filter and down-sample
>to get the true 48kHz signal.

Why not just use 48?

>Beats can get lost with low sample rates, so this may be important for
>some uses of alternative tunings.

How's that?

>> I can send you some papers off-list if you're interested in
>> the finer points of this topic.
>
>I swear I could hear differences in my Kyma distortion up to 64 kHz. I
>didn't try down-sampling it. There may have been a placebo effect.

64Khz recorded audio, or the signal path of a synthesizer being
sampled at 64KHz?

>Ambisonics is another thing to look at. I haven't, seriously, because I
>don't have the speakers.

You can get "super stereo" from certain kinds of Ambisonic recordings.

-Carl

Carl Lumma wrote:
> At 08:41 PM 12/19/2006, you wrote:

>> 96kHz sampling has a valid application for recording analog audio to be >> used at 48kHz. It means you can apply a digital filter and down-sample >> to get the true 48kHz signal.
> > Why not just use 48?

Because frequencies above 24kHz will be reflected into the audible range. You can get better results using a weak filter with 96, and then a strong, digital filter on that recording.

>> Beats can get lost with low sample rates, so this may be important for >> some uses of alternative tunings.
> > How's that?

You could be hearing beats between inaudible frequencies. Some tunings are designed so that the beats support the notes.

>>> I can send you some papers off-list if you're interested in
>>> the finer points of this topic.
>> I swear I could hear differences in my Kyma distortion up to 64 kHz. I >> didn't try down-sampling it. There may have been a placebo effect.
> > 64Khz recorded audio, or the signal path of a synthesizer being
> sampled at 64KHz?

64kHz processed audio, from my guitar.

>> Ambisonics is another thing to look at. I haven't, seriously, because I >> don't have the speakers.
> > You can get "super stereo" from certain kinds of Ambisonic recordings.

Yes, but you need different mixes for headphones or speakers. And it helps to make reverb part of the model. For now I'm using a simple panning/phase shifting model with a standard reverb. And I currently don't even have decent *stereo* speakers!

Graham

>>> 96kHz sampling has a valid application for recording analog audio to be
>>> used at 48kHz. It means you can apply a digital filter and down-sample
>>> to get the true 48kHz signal.
>>
>> Why not just use 48?
>
>Because frequencies above 24kHz will be reflected into the audible
>range. You can get better results using a weak filter with 96, and then
>a strong, digital filter on that recording.

Yes, I understand this, but I thought it was solved, even in
standard CD players, by oversampling in the playback D/A.

>>> Beats can get lost with low sample rates, so this may be important for
>>> some uses of alternative tunings.
>>
>> How's that?
>
>You could be hearing beats between inaudible frequencies.

These would be amplitude changes at the (inaudible) average of the
two inaudible frequencies.

>And I currently don't even have decent *stereo* speakers!

Therein lies the rub. Most people can't reproduce ultrasonics
even if they're in your recording.

-Carl

Carl Lumma wrote:
>>>> 96kHz sampling has a valid application for recording analog audio to be >>>> used at 48kHz. It means you can apply a digital filter and down-sample >>>> to get the true 48kHz signal.
>>> Why not just use 48?
>> Because frequencies above 24kHz will be reflected into the audible >> range. You can get better results using a weak filter with 96, and then >> a strong, digital filter on that recording.
> > Yes, I understand this, but I thought it was solved, even in
> standard CD players, by oversampling in the playback D/A.

No. You need filters on the inputs, to keep out erroneous frequencies, and filters on the outputs, to stop erroneous frequencies being introduced. The papers you gave support this. See the middle of page 9 of Stuart for a clear statement.

> >>>> Beats can get lost with low sample rates, so this may be important for >>>> some uses of alternative tunings.
>>> How's that?
>> You could be hearing beats between inaudible frequencies.
> > These would be amplitude changes at the (inaudible) average of the
> two inaudible frequencies.

I've heard reports that the result is audible -- I can't say how credible.

>> And I currently don't even have decent *stereo* speakers!
> > Therein lies the rub. Most people can't reproduce ultrasonics
> even if they're in your recording.

Yes, and such people wouldn't be looking for super-CD quality, so the whole thread becomes redundant.

Graham

On Wed, 20 Dec 2006, Graham Breed wrote:

> Csound uses floating point numbers for audio, which is even better for
> mastering because you don't have to worry about maximizing the levels.
> You can record audio files in floating point format. There are optional
> 64-bit builds as well. If you can hear the difference over 32 bits
> (some claim to) it's probably because of bugs in the system. But it
> doesn't slow things down much.

It is only when there is feedback in the system that 64 bits sounds better than 32 bits, in my experience. Specifically some digital filters can sound bad in the bass range if they use 32 bit state variables. So I think most of the chain can be 32 bits and only sensitive stuff needs to be in 64 bits.

/ Magnus Jonsson

>>> Because frequencies above 24kHz will be reflected into the audible
>>> range. You can get better results using a weak filter with 96, and
>>> then a strong, digital filter on that recording.
>>
>> Yes, I understand this, but I thought it was solved, even in
>> standard CD players, by oversampling in the playback D/A.
>
>No. You need filters on the inputs, to keep out erroneous frequencies,
>and filters on the outputs, to stop erroneous frequencies being
>introduced. The papers you gave support this.

Yes, I realized this right after I posted, but I wanted to finish
collecting/reading papers before I replied.

It looks like Stuart gives the best overall treatment of the topic
I've seen. Some of his graphs are hard to read, though. The critical
graph, number 25 -- if the red line is the human hearing threshold,
the cyan rectangle he's advocating hardly contains it.

I tend to disfavor the idea of relying on noise shaping -- we're not
*that* hard-pressed for storage space. So my approach would be, use
the smallest multiple of 44.1 (for easy conversion from CD) and the
smallest multiple of 8 (so it's a number of bytes) that contain the
threshold in a rectangle. It sounds like this would be 88.2 @ 24.
Except he says that with noise shaping, using a higher sampling rate
saves you bits/sample. But he doesn't say if this is also the case
for a rectangular channel.

He doesn't say how much precisely how much breathing room we need to
protect against various kinds of filter artifacts. For that, I have
only the word of dCs, who just happen to sell a $30K 196KHz.
upsampler box...

-Carl

> Super OT if you ask me.

Yeah, it's definitely getting there.

> I think formats of the music we produce is entirely on topic.

Well, if "formats of the music" were actually related to *microtonality*
somehow I'd be happier. As is, I'm just hitting the delete key on this
discussion.

But, I run a few mail lists, and my maxim for these lists is: "all
discussions annoy somebody"... :-)

Rick