Diterhing

[Carlo Centemeri]

Hello everybody,
I don't know if this is the right group where to post in, but I try
anyway.

I have the following situation: a commercial CD, which is playing a in
a very different way from CD player to CD player. From the symptoms
(slight detuning on high voice frequencies, occasional saturations) I
am supposing that the downgrading from the original 24 bit of the
master to the 16 bit of the CD has been done in a very bad way (and so
bits have been truncated, without doing a real dithering).
(I've had access to the original 24bit recording master and the
aforementioned problems do not exist at all)

My questions are:
1. since the performance on various Hi-Fi equipments are QUITE
different, do you think possible that a D/A converter - always talking
about common commercial products, not about audiophiles or hi-end cd
players - can have such different performances in recostructing?
2. apart from a possible bad reconversion of the 24 bit master, which
could be the causes of possible harmonic distorsions/saturations and
similar stuff?

Thank you

Carlo

[Anahata]

On Thu, 08 Oct 2009 02:03:02 -0700, Carlo Centemeri wrote:

> I have the following situation: a commercial CD, which is playing a in a
> very different way from CD player to CD player. From the symptoms
> (slight detuning on high voice frequencies, occasional saturations) I am
> supposing that the downgrading from the original 24 bit of the master to
> the 16 bit of the CD has been done in a very bad way (and so bits have
> been truncated, without doing a real dithering).

More likely it's clipping and some of your CD players can't actually
handle 0dBFS. It's likely to be the playback equiment that's at fault,
and the CD happens to contain material that shows it up.

I don't see a mechnism where truncation instead of dithering would cause
different players to respond differently.

--
Anahata
==//== 01638 720444
http://www.treewind.co.uk ==//== http://www.myspace.com/maryanahata

[Laurence Payne[_2_]]

On Thu, 8 Oct 2009 02:03:02 -0700 (PDT), Carlo Centemeri
> wrote:

>Hello everybody,
>I don't know if this is the right group where to post in, but I try
>anyway.
>
>I have the following situation: a commercial CD, which is playing a in
>a very different way from CD player to CD player. From the symptoms
>(slight detuning on high voice frequencies, occasional saturations) I
>am supposing that the downgrading from the original 24 bit of the
>master to the 16 bit of the CD has been done in a very bad way (and so
>bits have been truncated, without doing a real dithering).
>(I've had access to the original 24bit recording master and the
>aforementioned problems do not exist at all)
>
>My questions are:
>1. since the performance on various Hi-Fi equipments are QUITE
>different, do you think possible that a D/A converter - always talking
>about common commercial products, not about audiophiles or hi-end cd
>players - can have such different performances in recostructing?
>2. apart from a possible bad reconversion of the 24 bit master, which
>could be the causes of possible harmonic distorsions/saturations and
>similar stuff?

Does this happen in soft bits of the recording, or loud bits? If the
recording has been over-compressed and is flatlining up against the
0dB ceiling, some players cope with this more elegantly than others.

Try ripping the track and re-burning a CD after taking the overall
level down a notch. Does it now sound the same on all players?

[Arny Krueger]

"anahata" > wrote in message
o.uk
> On Thu, 08 Oct 2009 02:03:02 -0700, Carlo Centemeri wrote:
>
>> I have the following situation: a commercial CD, which
>> is playing a in a very different way from CD player to
>> CD player.

What do you know about the actual technical performance of the players?

What are they?

>> From the symptoms (slight detuning on high
>> voice frequencies, occasional saturations) I am
>> supposing that the downgrading from the original 24 bit
>> of the master to the 16 bit of the CD has been done in a
>> very bad way (and so bits have been truncated, without
>> doing a real dithering).

Very unlikely.

> More likely it's clipping and some of your CD players
> can't actually handle 0dBFS. It's likely to be the
> playback equiment that's at fault, and the CD happens to
> contain material that shows it up.

> I don't see a mechanism where truncation instead of
> dithering would cause different players to respond
> differently.

At some point it would - say if you truncate 16 bits from the CD down to
less than 12 bits.

However, I know of no good or even mediocre players that can do that.

[[email protected]]

Laurence Payne > wrote:
: Try ripping the track and re-burning a CD after taking the overall
: level down a notch. Does it now sound the same on all players?

If the cause turns out to be a clipped waveform, you could also try
running it through clip restoration software such as the plugin for
CoolEdit. On rare occasions, that can help a little. :)

[Carlo Centemeri]

> Carlo,
> * * *Different converters do sound different. I have found that a modern
> hybrid converter sounds harsh on some older recordings, where A/D conversion
> was not as advanced. Single-bit converters, which were common up to about
> Y2K, sound smoother on such recordings. Technically, hybrid converters are
> better, but the lack of low level linearity inherent in a single bit
> converter can smooth out some recordings.
>
> Bob Morein
> (310) 237-6511

Dear Bob,
thank you (and thank everybody) for the answers.
I'll clarify a bit (just one, not 24 :-)..) my position: I am a long
time CD buyer and CD reviewer for several magazines, working
expecially on classical and baroque music. Being moreover a musician
myself and having a technical background, I have always tried to
understand the reasons of what my ears communicate me.

It is obvious that a 24 bit recording should sound better than a 16
bit (otherwise why recording at 16) and that different converters
sound different: numbers are numbers and if the output is different,
something must happen :-)

Said this: having listened to several (thousand of) cds, I have met
this recording in which, for the first time in my life, I have found
out something which from a player to another does not seem even the
same record (it seems indeed the same recording in different
editions). I am comparing player very different in between them, all
in the commercial sector (some rack equipment: a yamaha, a technics
and a Luxman, all three through the same Luxman ampli, and some
compact stereo equipment, a sony, a technics and a Sharp). All test
made with the same AKG earphone.
The order of appreciation I have is pretty strange, because for
example the sony compact sounds far more better than the yamaha rack
(!).

Moreover, I use frequently all those players, and no other record has
ever sounded more different (I tried with some other cds in the same
time, just to understand if my ears were getting crazy)

But apart from this, which could actually be the 0dbFS handling
compliance or not (and in fact - apart from the Luxman - the older
sound worse than the newer), what I was flashed about was the frequent
saturation present in the recording. I asked to listen to the original
24bit tracks and there is no saturation/distortion at all.

I can then modify my question in:

do you think is normal that downgrading from 24 to 16 bits all the
"LOUD" tracks of a certain recording start clamping?
or better
since no compressor was used, which could be the cause of a series of
clamp that originate during the process that goes from editing to
mastering?

Thank you really very much!

Carlo

[Wecan do it]

"Carlo Centemeri" > wrote in message
...
> Carlo,
> Different converters do sound different. I have found that a
> modern
> hybrid converter sounds harsh on some older recordings,
> where A/D conversion
> was not as advanced. Single-bit converters, which were
> common up to about
> Y2K, sound smoother on such recordings. Technically, hybrid
> converters are
> better, but the lack of low level linearity inherent in a
> single bit
> converter can smooth out some recordings.
>
> Bob Morein
> (310) 237-6511

Dear Bob,
thank you (and thank everybody) for the answers.
I'll clarify a bit (just one, not 24 :-)..) my position: I am
a long
time CD buyer and CD reviewer for several magazines, working
expecially on classical and baroque music. Being moreover a
musician
myself and having a technical background, I have always tried
to
understand the reasons of what my ears communicate me.

It is obvious that a 24 bit recording should sound better than
a 16
bit (otherwise why recording at 16) and that different
converters
sound different: numbers are numbers and if the output is
different,
something must happen :-)

Said this: having listened to several (thousand of) cds, I
have met
this recording in which, for the first time in my life, I have
found
out something which from a player to another does not seem
even the
same record (it seems indeed the same recording in different
editions). I am comparing player very different in between
them, all
in the commercial sector (some rack equipment: a yamaha, a
technics
and a Luxman, all three through the same Luxman ampli, and
some
compact stereo equipment, a sony, a technics and a Sharp). All
test
made with the same AKG earphone.
The order of appreciation I have is pretty strange, because
for
example the sony compact sounds far more better than the
yamaha rack
(!).

Moreover, I use frequently all those players, and no other
record has
ever sounded more different (I tried with some other cds in
the same
time, just to understand if my ears were getting crazy)

But apart from this, which could actually be the 0dbFS
handling
compliance or not (and in fact - apart from the Luxman - the
older
sound worse than the newer), what I was flashed about was the
frequent
saturation present in the recording. I asked to listen to the
original
24bit tracks and there is no saturation/distortion at all.

I can then modify my question in:

do you think is normal that downgrading from 24 to 16 bits all
the
"LOUD" tracks of a certain recording start clamping?
or better
since no compressor was used, which could be the cause of a
series of
clamp that originate during the process that goes from editing
to
mastering?

Thank you really very much!

Carlo
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

The on thing consistant between 8 or 16 or 32 bit recordings
is that if they are all ones you get 0dBfsd. No more no less.
Neither can be louder than the other. Even during dithering
down to from higher to lower bit rate.

My bet is the disk is marginal and one player plugs in a lot
more error correction during playback than another does. I
have players like this I use to test disks I believe are
marginal.


peace
dawg

[Carlo Centemeri]

> My bet is the disk is marginal and one player plugs in a lot
> more error correction during playback than another does. I
> have players like this I use to test disks I believe are
> marginal.

Sorry, when you talk about marginals you mean a disc with "physical"
handicaps (such as low reflectivity) etc, which then could cause a
high error rate due to hardly/badly readable bits?
Thank you
Carlo

[Ethan Winer[_3_]]

On Oct 8, 11:42 am, Carlo Centemeri > wrote:
> It is obvious that a 24 bit recording should sound better than a 16
> bit (otherwise why recording at 16) and that different converters
> sound different: numbers are numbers and if the output is different,
> something must happen :-)

What seems obvious is not always so. In my experience, 24 bits used as
a delivery medium has absolutely no advantage over 16 bits. Further,
Meyer and Moran pretty well proved the point in their AES Journal
article described here:

http://mixonline.com/recording/mixing/audio_emperors_new_sampling/index.html

Same for dither. Propeller-heads can "prove" that dither reduces
distortion - and it does! - but the improvement is not audible. More
here:

http://www.ethanwiner.com/dither.html

BTW, I use "propeller-heads" in the most complimentary manner because
I are one too. :->)

--Ethan

[Mark]

On Oct 8, 12:50*pm, Ethan Winer > wrote:
> On Oct 8, 11:42 am, Carlo Centemeri > wrote:
>
> > It is obvious that a 24 bit recording should sound better than a 16
> > bit (otherwise why recording at 16) and that different converters
> > sound different: numbers are numbers and if the output is different,
> > something must happen :-)
>
> What seems obvious is not always so. In my experience, 24 bits used as
> a delivery medium has absolutely no advantage over 16 bits. Further,
> Meyer and Moran pretty well proved the point in their AES Journal
> article described here:
>
> http://mixonline.com/recording/mixing/audio_emperors_new_sampling/ind...
>
> Same for dither. Propeller-heads can "prove" that dither reduces
> distortion - and it does! - but the improvement is not audible. More
> here:
>
> http://www.ethanwiner.com/dither.html
>
> BTW, I use "propeller-heads" in the most complimentary manner because
> I are one too. :->)
>
> --Ethan

Ethan,

I agree with you that the improvement due to proper dither on a 16 bit
recording, while it certainly IS measureable, may not be audible.

So I suggest an addition to your demo...create another comparison, but
instead of using 16 bits, reduce the bit depth to 8 or 10 or whatever
it takes to make the difference more obvious. It would be interesting
to listen to. Once you have trained your ear to hear the difference
at 8 or 10 bits, it would be interesting to go back and listen to the
16 bit demo again.

Mark

[Laurence Payne[_2_]]

On Thu, 8 Oct 2009 08:42:46 -0700 (PDT), Carlo Centemeri
> wrote:

>I can then modify my question in:
>
>do you think is normal that downgrading from 24 to 16 bits all the
>"LOUD" tracks of a certain recording start clamping?
>or better
>since no compressor was used, which could be the cause of a series of
>clamp that originate during the process that goes from editing to
>mastering?

I think you're chasing the wrong horse. 16 bit playback includes all
the information your ears can handle, and more. 24 bit has advantages
when recording and mixing/processing but doesn't sound "better".

Have you tried another sample of the same CD? Maybe you've just got a
really bad pressing, and some players are coping with an enormous
demand for error-correction better than others.

[Scott Dorsey]

Carlo Centemeri > wrote:
>
>I have the following situation: a commercial CD, which is playing a in
>a very different way from CD player to CD player. From the symptoms
>(slight detuning on high voice frequencies, occasional saturations) I
>am supposing that the downgrading from the original 24 bit of the
>master to the 16 bit of the CD has been done in a very bad way (and so
>bits have been truncated, without doing a real dithering).
>(I've had access to the original 24bit recording master and the
>aforementioned problems do not exist at all)

No. What you are hearing are errors. CDs have a very high error rate
but also very good error correction. However, some CDs have a higher error
rate than others and some players have better error correction than others.

Hold the disc up to the light and see if you see any pinholes in it.
Metallization problems are the number one source of these issues.

Barring that, use a computer CD-R drive... load in a disc image, and dump
it back out. The CD-R drive can reread when it encounters soft errors,
which audio drives cannot do. You should get a copy that plays well.

The issue has nothing to do with dithering.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Ethan Winer[_3_]]

On Oct 8, 2:27 pm, Mark > wrote:
> So I suggest an addition to your demo...create another comparison, but
> instead of using 16 bits, reduce the bit depth to 8 or 10 or whatever
> it takes to make the difference more obvious.

Way ahead of you Mark. :->)

Another of my demos will use a bit-reducer plug-in to reduce the bit
depth of a recording in single bit steps. The track is a cello
orchestra playing Chic Corea's Spain, so it's a good source for this
test. The playback is pretty clean until you get down to 10 or 11
bits! But the plug-in only reduces the bits, it doesn't also add
dither.

--Ethan

[Don Pearce[_3_]]

On Mon, 12 Oct 2009 05:14:05 -0700 (PDT), Ethan Winer
> wrote:

>On Oct 8, 2:27 pm, Mark > wrote:
>> So I suggest an addition to your demo...create another comparison, but
>> instead of using 16 bits, reduce the bit depth to 8 or 10 or whatever
>> it takes to make the difference more obvious.
>
>Way ahead of you Mark. :->)
>
>Another of my demos will use a bit-reducer plug-in to reduce the bit
>depth of a recording in single bit steps. The track is a cello
>orchestra playing Chic Corea's Spain, so it's a good source for this
>test. The playback is pretty clean until you get down to 10 or 11
>bits! But the plug-in only reduces the bits, it doesn't also add
>dither.
>
Unfair doing it without dither. NICAM on the UK analogue TV service is
10 bits (or the most significant necessary 10 from 14 during a 1mSec
window) and sound just fine.

d

[Mike Rivers[_2_]]

Don Pearce wrote:

> Unfair doing it without dither. NICAM on the UK analogue TV service is
> 10 bits (or the most significant necessary 10 from 14 during a 1mSec
> window) and sound just fine.

It's not unfair, it's just a demonstration of what fewer bits of
resolution sounds
like. It may help certain listeners understand when it's important to
work with
24-bit recordings and when it doesn't really matter.

Add dither after listening and you can hear what's usually
considered an improvement made by dithering. And if your
listening environment is good enough you might also hear the
added noise.

[Don Pearce[_3_]]

On Mon, 12 Oct 2009 09:17:09 -0400, Mike Rivers >
wrote:

>Don Pearce wrote:
>
>> Unfair doing it without dither. NICAM on the UK analogue TV service is
>> 10 bits (or the most significant necessary 10 from 14 during a 1mSec
>> window) and sound just fine.
>
>It's not unfair, it's just a demonstration of what fewer bits of
>resolution sounds

Oh I have to disagree. Dithering is an essential and integral part of
digitizing audio (or anything for that matter), and to present a
sample without it doesn't give a demonstration of what it sounds like.

>like. It may help certain listeners understand when it's important to
>work with
>24-bit recordings and when it doesn't really matter.
>

Not at all. For a start it is impossible to make a 24 bit recording
without dither. And of course in the real world 24 bits gives you a
bare 10dB S/N advantage over 16 bits, and I don't believe there is a
single real-world situation (outside an ultra quiet lab) where 24 bits
are either necessary or discernibly better than 16.

>Add dither after listening and you can hear what's usually
>considered an improvement made by dithering. And if your
>listening environment is good enough you might also hear the
>added noise.

No, the noise is much less intrusive than a "foreign" tone present
alongside the wanted ones. Add shaping to the noise to skew the
spectrum up towards the ultrasonic, and it is so much the better.

d

[Johann Spischak]

"Mike Rivers" > schrieb im Newsbeitrag
...
> Don Pearce wrote:
>
>> Unfair doing it without dither. NICAM on the UK analogue TV service is
>> 10 bits (or the most significant necessary 10 from 14 during a 1mSec
>> window) and sound just fine.
>
> It's not unfair, it's just a demonstration of what fewer bits of
> resolution sounds
> like. It may help certain listeners understand when it's important to work
> with
> 24-bit recordings and when it doesn't really matter.
>
> Add dither after listening and you can hear what's usually
> considered an improvement made by dithering. And if your
> listening environment is good enough you might also hear the
> added noise.


Hello Mike, I can confirm, that over a specific quality level the caused
damages becoming more and more hearable. By really good hearing and
digital-converting conditions, it is totally unusable. You will remember,
about two years ago we have discussed here this theme in combination with
noise and distortion problems in digital sampling.

I am describing dither always with a scene when we passing a garden fence
by. According to our speed we can see more or less behind it. Over a
specific speed (16 bit/ 44 kHz) it is better to take that damned fence away
:-)))

Best regards, Johann

[Don Pearce[_3_]]

On Tue, 13 Oct 2009 12:28:48 +0200, "Johann Spischak"
> wrote:

>"Mike Rivers" > schrieb im Newsbeitrag
...
>> Don Pearce wrote:
>>
>>> Unfair doing it without dither. NICAM on the UK analogue TV service is
>>> 10 bits (or the most significant necessary 10 from 14 during a 1mSec
>>> window) and sound just fine.
>>
>> It's not unfair, it's just a demonstration of what fewer bits of
>> resolution sounds
>> like. It may help certain listeners understand when it's important to work
>> with
>> 24-bit recordings and when it doesn't really matter.
>>
>> Add dither after listening and you can hear what's usually
>> considered an improvement made by dithering. And if your
>> listening environment is good enough you might also hear the
>> added noise.
>
>
>Hello Mike, I can confirm, that over a specific quality level the caused
>damages becoming more and more hearable. By really good hearing and
>digital-converting conditions, it is totally unusable. You will remember,
>about two years ago we have discussed here this theme in combination with
>noise and distortion problems in digital sampling.
>
>I am describing dither always with a scene when we passing a garden fence
>by. According to our speed we can see more or less behind it. Over a
>specific speed (16 bit/ 44 kHz) it is better to take that damned fence away
>:-)))
>
>Best regards, Johann

Even allowing for English being your second language, I can make no
sense of this, I'm afraid. Do you think you could explain again, but
this time stating exactly what you are talking about, rather than just
using words like "it", which could mean anything?

d

[Johann Spischak]

>>Hello Mike, I can confirm, that over a specific quality level the caused
>>damages becoming more and more hearable. By really good hearing and
>>digital-converting conditions, it is totally unusable. You will remember,
>>about two years ago we have discussed here this theme in combination with
>>noise and distortion problems in digital sampling.
>>
>>I am describing dither always with a scene when we passing a garden fence
>>by. According to our speed we can see more or less behind it. Over a
>>specific speed (16 bit/ 44 kHz) it is better to take that damned fence
>>away
>>:-)))
>>
>>Best regards, Johann
>
> Even allowing for English being your second language, I can make no
> sense of this, I'm afraid. Do you think you could explain again, but
> this time stating exactly what you are talking about, rather than just
> using words like "it", which could mean anything?
>
> d

Of course, your welcome. I'm talking about what happens when you passing a
picket fence by. You need only to imagine "it". :-))

regads, Johann

[Don Pearce[_3_]]

On Tue, 13 Oct 2009 13:10:38 +0200, "Johann Spischak"
> wrote:

>>>Hello Mike, I can confirm, that over a specific quality level the caused
>>>damages becoming more and more hearable. By really good hearing and
>>>digital-converting conditions, it is totally unusable. You will remember,
>>>about two years ago we have discussed here this theme in combination with
>>>noise and distortion problems in digital sampling.
>>>
>>>I am describing dither always with a scene when we passing a garden fence
>>>by. According to our speed we can see more or less behind it. Over a
>>>specific speed (16 bit/ 44 kHz) it is better to take that damned fence
>>>away
>>>:-)))
>>>
>>>Best regards, Johann
>>
>> Even allowing for English being your second language, I can make no
>> sense of this, I'm afraid. Do you think you could explain again, but
>> this time stating exactly what you are talking about, rather than just
>> using words like "it", which could mean anything?
>>
>> d
>
>Of course, your welcome. I'm talking about what happens when you passing a
>picket fence by. You need only to imagine "it". :-))
>
>regads, Johann

That is less explanation, not more. I would like to understand both
your paragraphs, but I can't follow what you are saying in either of
them.

d

[Johann Spischak]

"Don Pearce" > schrieb im Newsbeitrag
...
> On Tue, 13 Oct 2009 13:10:38 +0200, "Johann Spischak"
> > wrote:
>
>>>>Hello Mike, I can confirm, that over a specific quality level the caused
>>>>damages becoming more and more hearable. By really good hearing and
>>>>digital-converting conditions, it is totally unusable. You will
>>>>remember,
>>>>about two years ago we have discussed here this theme in combination
>>>>with
>>>>noise and distortion problems in digital sampling.
>>>>
>>>>I am describing dither always with a scene when we passing a garden
>>>>fence
>>>>by. According to our speed we can see more or less behind it. Over a
>>>>specific speed (16 bit/ 44 kHz) it is better to take that damned fence
>>>>away
>>>>:-)))
>>>>
>>>>Best regards, Johann
>>>
>>> Even allowing for English being your second language, I can make no
>>> sense of this, I'm afraid. Do you think you could explain again, but
>>> this time stating exactly what you are talking about, rather than just
>>> using words like "it", which could mean anything?
>>>
>>> d
>>
>>Of course, your welcome. I'm talking about what happens when you passing a
>>picket fence by. You need only to imagine "it". :-))
>>
>>regads, Johann
>
> That is less explanation, not more. I would like to understand both
> your paragraphs, but I can't follow what you are saying in either of
> them.
>
> d

Well, if you don't understand the first paragraph, then you need to ask
Mike, I have only confirmed, what he wrote. For the second, you must be
aware of how dithering works, then with a little phantasy you will see, that
dithering is like looking through the gaps between the pickets through. By
staying - walking - running. :-))

Regards, Johann

[Don Pearce[_3_]]

On Tue, 13 Oct 2009 13:51:58 +0200, "Johann Spischak"
> wrote:

>"Don Pearce" > schrieb im Newsbeitrag
...
>> On Tue, 13 Oct 2009 13:10:38 +0200, "Johann Spischak"
>> > wrote:
>>
>>>>>Hello Mike, I can confirm, that over a specific quality level the caused
>>>>>damages becoming more and more hearable. By really good hearing and
>>>>>digital-converting conditions, it is totally unusable. You will
>>>>>remember,
>>>>>about two years ago we have discussed here this theme in combination
>>>>>with
>>>>>noise and distortion problems in digital sampling.
>>>>>
>>>>>I am describing dither always with a scene when we passing a garden
>>>>>fence
>>>>>by. According to our speed we can see more or less behind it. Over a
>>>>>specific speed (16 bit/ 44 kHz) it is better to take that damned fence
>>>>>away
>>>>>:-)))
>>>>>
>>>>>Best regards, Johann
>>>>
>>>> Even allowing for English being your second language, I can make no
>>>> sense of this, I'm afraid. Do you think you could explain again, but
>>>> this time stating exactly what you are talking about, rather than just
>>>> using words like "it", which could mean anything?
>>>>
>>>> d
>>>
>>>Of course, your welcome. I'm talking about what happens when you passing a
>>>picket fence by. You need only to imagine "it". :-))
>>>
>>>regads, Johann
>>
>> That is less explanation, not more. I would like to understand both
>> your paragraphs, but I can't follow what you are saying in either of
>> them.
>>
>> d
>
>Well, if you don't understand the first paragraph, then you need to ask
>Mike, I have only confirmed, what he wrote. For the second, you must be
>aware of how dithering works, then with a little phantasy you will see, that
>dithering is like looking through the gaps between the pickets through. By
>staying - walking - running. :-))
>
>Regards, Johann

As regards the first, ok.

For the second, no, that is not dithering, it is sampling which is an
entirely different matter. Moving past at different speeds is the same
as changing the sampling rate. There is no equivalent to dithering
possible in this analogy.

d

[Johann Spischak]

"Don Pearce" > schrieb im Newsbeitrag
...
> On Tue, 13 Oct 2009 13:51:58 +0200, "Johann Spischak"
> > wrote:
>
>>"Don Pearce" > schrieb im Newsbeitrag
...
>>> On Tue, 13 Oct 2009 13:10:38 +0200, "Johann Spischak"
>>> > wrote:
>>>
>>>>>>Hello Mike, I can confirm, that over a specific quality level the
>>>>>>caused
>>>>>>damages becoming more and more hearable. By really good hearing and
>>>>>>digital-converting conditions, it is totally unusable. You will
>>>>>>remember,
>>>>>>about two years ago we have discussed here this theme in combination
>>>>>>with
>>>>>>noise and distortion problems in digital sampling.
>>>>>>
>>>>>>I am describing dither always with a scene when we passing a garden
>>>>>>fence
>>>>>>by. According to our speed we can see more or less behind it. Over a
>>>>>>specific speed (16 bit/ 44 kHz) it is better to take that damned fence
>>>>>>away
>>>>>>:-)))
>>>>>>
>>>>>>Best regards, Johann
>>>>>
>>>>> Even allowing for English being your second language, I can make no
>>>>> sense of this, I'm afraid. Do you think you could explain again, but
>>>>> this time stating exactly what you are talking about, rather than just
>>>>> using words like "it", which could mean anything?
>>>>>
>>>>> d
>>>>
>>>>Of course, your welcome. I'm talking about what happens when you passing
>>>>a
>>>>picket fence by. You need only to imagine "it". :-))
>>>>
>>>>regads, Johann
>>>
>>> That is less explanation, not more. I would like to understand both
>>> your paragraphs, but I can't follow what you are saying in either of
>>> them.
>>>
>>> d
>>
>>Well, if you don't understand the first paragraph, then you need to ask
>>Mike, I have only confirmed, what he wrote. For the second, you must be
>>aware of how dithering works, then with a little phantasy you will see,
>>that
>>dithering is like looking through the gaps between the pickets through. By
>>staying - walking - running. :-))
>>
>>Regards, Johann
>
> As regards the first, ok.
>
> For the second, no, that is not dithering, it is sampling which is an
> entirely different matter. Moving past at different speeds is the same
> as changing the sampling rate. There is no equivalent to dithering
> possible in this analogy.
>
> d

Well, if you think on how dithering works, what it does with which bits and
how, you wil see, the picket fence is the dither itself.

Regards, Johann

[William Sommerwerck]

I don't see the connection between sampling and picket fences. Moving past a
stationary fence really isn't sampling, because the movement isn't
time-quantized. You actually see "everything" behind the fence, not just
samples.

As for dithering... my understanding is that its purpose is, by adding
noise, to turn the "correlated" quantization errors that occur with
repetitive signals into uncorrelated and more-noise-like errors. Again, I
don't see a connection with picket fences.

[Don Pearce[_3_]]

On Tue, 13 Oct 2009 14:44:07 +0200, "Johann Spischak"
> wrote:

>"Don Pearce" > schrieb im Newsbeitrag
...
>> On Tue, 13 Oct 2009 13:51:58 +0200, "Johann Spischak"
>> > wrote:
>>
>>>"Don Pearce" > schrieb im Newsbeitrag
...
>>>> On Tue, 13 Oct 2009 13:10:38 +0200, "Johann Spischak"
>>>> > wrote:
>>>>
>>>>>>>Hello Mike, I can confirm, that over a specific quality level the
>>>>>>>caused
>>>>>>>damages becoming more and more hearable. By really good hearing and
>>>>>>>digital-converting conditions, it is totally unusable. You will
>>>>>>>remember,
>>>>>>>about two years ago we have discussed here this theme in combination
>>>>>>>with
>>>>>>>noise and distortion problems in digital sampling.
>>>>>>>
>>>>>>>I am describing dither always with a scene when we passing a garden
>>>>>>>fence
>>>>>>>by. According to our speed we can see more or less behind it. Over a
>>>>>>>specific speed (16 bit/ 44 kHz) it is better to take that damned fence
>>>>>>>away
>>>>>>>:-)))
>>>>>>>
>>>>>>>Best regards, Johann
>>>>>>
>>>>>> Even allowing for English being your second language, I can make no
>>>>>> sense of this, I'm afraid. Do you think you could explain again, but
>>>>>> this time stating exactly what you are talking about, rather than just
>>>>>> using words like "it", which could mean anything?
>>>>>>
>>>>>> d
>>>>>
>>>>>Of course, your welcome. I'm talking about what happens when you passing
>>>>>a
>>>>>picket fence by. You need only to imagine "it". :-))
>>>>>
>>>>>regads, Johann
>>>>
>>>> That is less explanation, not more. I would like to understand both
>>>> your paragraphs, but I can't follow what you are saying in either of
>>>> them.
>>>>
>>>> d
>>>
>>>Well, if you don't understand the first paragraph, then you need to ask
>>>Mike, I have only confirmed, what he wrote. For the second, you must be
>>>aware of how dithering works, then with a little phantasy you will see,
>>>that
>>>dithering is like looking through the gaps between the pickets through. By
>>>staying - walking - running. :-))
>>>
>>>Regards, Johann
>>
>> As regards the first, ok.
>>
>> For the second, no, that is not dithering, it is sampling which is an
>> entirely different matter. Moving past at different speeds is the same
>> as changing the sampling rate. There is no equivalent to dithering
>> possible in this analogy.
>>
>> d
>
>Well, if you think on how dithering works, what it does with which bits and
>how, you wil see, the picket fence is the dither itself.
>
>Regards, Johann

No, the picket fence is the sampler.

d

[Mike Rivers]

Don Pearce wrote:

> Oh I have to disagree. Dithering is an essential and integral part of
> digitizing audio (or anything for that matter), and to present a
> sample without it doesn't give a demonstration of what it sounds like.

I didn't mean to imply that dithering wasn't important. But if you want to
demonstrate how it helps, it's possible to concoct a demonstration with
and without added dither that does so.

> Not at all. For a start it is impossible to make a 24 bit recording
> without dither.

That's kind of a loaded statement. There is, at least within the realm
of practicality always a certain amount of random noise at greater
amplitude than the lowest order bit in any "high resolution" digital
recording system. So there will always be dither. But what we're
talking about here (at least what I'm talking about, and I think what
Ethan's talking about) is ADDED dither, some controlled random
noise with a known probability density.

> And of course in the real world 24 bits gives you a
> bare 10dB S/N advantage over 16 bits, and I don't believe there is a
> single real-world situation (outside an ultra quiet lab) where 24 bits
> are either necessary or discernibly better than 16.

It depends on how you use those bits. I use them to give myself
more protection from clipping when I can't fully control the signal
level. I can take advantage of the additional 10 dB S/N by keeping
my controllable peaks 10 dB lower than I might otherwise do with
16-bit recording.

>> Add dither after listening and you can hear what's usually
>> considered an improvement made by dithering. And if your
>> listening environment is good enough you might also hear the
>> added noise.

> No, the noise is much less intrusive than a "foreign" tone present
> alongside the wanted ones.

Certainly. But if you want to hear the dither (that is, the noise that is
applied for the purpose of assuring that the sighal is dithered) it is
sometimes possible. This has nothing to do with the EFFECT of the
dither on the desired signal.

Surely you know the classic experiment. Record a piano note at
-65 dBFS peak using 16-bit resolution and no added dither. Crank
up the playback level by 65 dB and listen to it. The granularity of
the playback will be obvious on the decay.

Now, record it at 24-bit resolution, at the same level, apply dither,
and truncate it to 16 bits. Again, jack up the playback level and listen.
It will be just as noisy as the straight 16-bit recording, but the decay
will be much smoother.

This is an illustration of the benefit of adding dither. It can be extended
by adding dither to the 16-bit recording. That will raise the noise level,
but the decay into the noise will be smoother as a result of the dithering.

[Ethan Winer[_3_]]

On Oct 12, 9:17 am, Mike Rivers > wrote:
> It's not unfair, it's just a demonstration of what fewer bits of
> resolution sounds like.

Exactly. If anything, not using dither works against me for calling BS
on those who say 16 bits is not enough. Hey, if 13 bits sounds fine
WITHOUT dither, then imagine how much better it will sound with
dither.

Alas, I never got to that part of the demo. :-(

A friend and I video taped the entire workshop using two cameras. Over
the next few weeks I plan to make a video of the best parts, and I'll
probably add to the video some of the demos I didn't have time for at
the presentation.

--Ethan

[Don Pearce]

On Tue, 13 Oct 2009 09:35:47 -0400, Mike Rivers >
wrote:

>Don Pearce wrote:
>
>> Oh I have to disagree. Dithering is an essential and integral part of
>> digitizing audio (or anything for that matter), and to present a
>> sample without it doesn't give a demonstration of what it sounds like.
>
>I didn't mean to imply that dithering wasn't important. But if you want to
>demonstrate how it helps, it's possible to concoct a demonstration with
>and without added dither that does so.
>
Ah, sure yes I get that.

>> Not at all. For a start it is impossible to make a 24 bit recording
>> without dither.
>
>That's kind of a loaded statement. There is, at least within the realm
>of practicality always a certain amount of random noise at greater
>amplitude than the lowest order bit in any "high resolution" digital
>recording system. So there will always be dither. But what we're
>talking about here (at least what I'm talking about, and I think what
>Ethan's talking about) is ADDED dither, some controlled random
>noise with a known probability density.
>

Dither is obviously the sum of some specific addition and whatever
noise is already there. At 24 bits, the random accidental stuff is
overwhelmingly bigger than anything added. in fact the added stuff
will be essentially invisible, so maybe a loaded statement, but also a
perfectly true one.

>> And of course in the real world 24 bits gives you a
>> bare 10dB S/N advantage over 16 bits, and I don't believe there is a
>> single real-world situation (outside an ultra quiet lab) where 24 bits
>> are either necessary or discernibly better than 16.
>
>It depends on how you use those bits. I use them to give myself
>more protection from clipping when I can't fully control the signal
>level. I can take advantage of the additional 10 dB S/N by keeping
>my controllable peaks 10 dB lower than I might otherwise do with
>16-bit recording.
>
>>> Add dither after listening and you can hear what's usually
>>> considered an improvement made by dithering. And if your
>>> listening environment is good enough you might also hear the
>>> added noise.
>
>> No, the noise is much less intrusive than a "foreign" tone present
>> alongside the wanted ones.
>
>Certainly. But if you want to hear the dither (that is, the noise that is
>applied for the purpose of assuring that the sighal is dithered) it is
>sometimes possible. This has nothing to do with the EFFECT of the
>dither on the desired signal.
>
Ok, almost different things but still inextricably linked.

>Surely you know the classic experiment. Record a piano note at
>-65 dBFS peak using 16-bit resolution and no added dither. Crank
>up the playback level by 65 dB and listen to it. The granularity of
>the playback will be obvious on the decay.
>
>Now, record it at 24-bit resolution, at the same level, apply dither,
>and truncate it to 16 bits. Again, jack up the playback level and listen.
>It will be just as noisy as the straight 16-bit recording, but the decay
>will be much smoother.
>
Yes I have done this. Of course if you are recording at -65dB you have
a 4 bit, not 16 bit recording. Even a phone is far better than that.
As a demonstration of quantization error it is obviously readily
audible, but so what? Record the same piano decay to tape at -65dB and
see what you can hear - that is even more instructive.

>This is an illustration of the benefit of adding dither. It can be extended
>by adding dither to the 16-bit recording. That will raise the noise level,
>but the decay into the noise will be smoother as a result of the dithering.

Yes.

d

[Mike Rivers]

Don Pearce wrote:

> Dither is obviously the sum of some specific addition and whatever
> noise is already there. At 24 bits, the random accidental stuff is
> overwhelmingly bigger than anything added. in fact the added stuff
> will be essentially invisible, so maybe a loaded statement, but also a
> perfectly true one.

I suppose that to continue this discussion we'd have to decide where
the dithered audio was going next. If it was a 24-bit word that was going
to a 16-bit guillotine, you'd want to apply dither at a level that would
make at least the 16th bit toggle randomly so that it could be painlessly
truncated. With a really good converter, you might have 3 or 4 bits of
random noise so you need to make up the rest of the 8 or 9 bits with
dither that you add.

> Of course if you are recording at -65dB you have
> a 4 bit, not 16 bit recording. Even a phone is far better than that.

This is why it's a concocted experiment to hear the improvement that
dither can make.

> As a demonstration of quantization error it is obviously readily
> audible, but so what? Record the same piano decay to tape at -65dB and
> see what you can hear - that is even more instructive.

But we're not doing that. Nobody ever said that a demonstration of a
principle has to be practical. You should have seen the laser-and-smoke
microphone at the AES show last weekend.

[Geoff]

I just can never decide if to dither, or not.....


geoff

[Richard Crowley]

"geoff" wrote ...
>I just can never decide if to dither, or not.....

See there? You're DITHERING now!
When in doubt, dither.

[Randy Yates]

"geoff" > writes:

> I just can never decide if to dither, or not.....

Better than being a dithering idiot...
--
Randy Yates % "Watching all the days go by...
Digital Signal Labs % Who are you and who am I?"
% 'Mission (A World Record)',
http://www.digitalsignallabs.com % *A New World Record*, ELO

[Anahata]

On Tue, 13 Oct 2009 13:04:36 +0000, Don Pearce wrote:

> No, the picket fence is the sampler.

Actually the only way a fence would be an analogy of a sampler is if you
are static and the slits in the fence are sufficiently narrowly spaced to
capture and reconstruct as much of the detail behind as you want to see.

Even moving the fence from side to side a little and taking an average
for each slit still wouldn't be analogous to dither - but it would be the
equivalent of an antialiasing filter.


--
Anahata
==//== 01638 720444
http://www.treewind.co.uk

[Arkansan Raider]

Randy Yates wrote:
> "geoff" > writes:
>
>> I just can never decide if to dither, or not.....
>
> Better than being a dithering idiot...

ROTFLOL

+1

That was punny. (=^D


---Jeff

[Arny Krueger]

"Don Pearce" > wrote in message

> On Mon, 12 Oct 2009 05:14:05 -0700 (PDT), Ethan Winer
> > wrote:
>
>> On Oct 8, 2:27 pm, Mark > wrote:
>>> So I suggest an addition to your demo...create another
>>> comparison, but instead of using 16 bits, reduce the
>>> bit depth to 8 or 10 or whatever it takes to make the
>>> difference more obvious.
>>
>> Way ahead of you Mark. :->)
>>
>> Another of my demos will use a bit-reducer plug-in to
>> reduce the bit depth of a recording in single bit steps.
>> The track is a cello orchestra playing Chic Corea's
>> Spain, so it's a good source for this test. The playback
>> is pretty clean until you get down to 10 or 11 bits! But
>> the plug-in only reduces the bits, it doesn't also add
>> dither.
>>
> Unfair doing it without dither. NICAM on the UK analogue
> TV service is 10 bits (or the most significant necessary
> 10 from 14 during a 1mSec window) and sound just fine.

I think that Ethan is trying to make the point that a lot of the heavy
breathing about dither is senseless.

Of course word truncation without dither is a very bad thing. Of course
dither should always be used at such times. Furthermore, as your example may
show, dither can make amazing amounts of very crude data reduction (by
truncation) remarkably palatable to the human ear.

On the other side we have the purveyors of golden dither who claim that
without the use of their product, every analog-to-digital conversion, even
at the 20 bit (actual) level, is grotesquely audibly flawed. I've been
public personally abused by some well-known people of this ilk, and
therefore have some problems being perfectly calm when they do their
deceptive little song-and-dance acts.

[Richard Crowley]

"Arny Krueger" wrote ...
> I think that Ethan is trying to make the point that a lot of the heavy
> breathing about dither is senseless.

Dither is not the only topic about which all the heavy
breathing is senseless.

[Arny Krueger]

"Richard Crowley" > wrote in message


> "Arny Krueger" wrote ...

>> I think that Ethan is trying to make the point that a
>> lot of the heavy breathing about dither is senseless.

> Dither is not the only topic about which all the heavy
> breathing is senseless.

Agreed. Another one relates to the alleged sonic problems of "brick wall
filters".