Dirty Digital [sic.]
 

In article , Rob
wrote:
Don Pearce wrote:



No. Dither is an integral part of the process - always has been and
always will be. It isn't shaping anything.

Yes, OK, I did say I was coming from a position of ignorance. I thought
dither does shape sound.

No. It allows the 'sound' to be recorded correctly.

Is dither a necessary

It (or an equivalent) is required if you want to avoid needlessly spoiling
the recording.

Just like with LP you have to get various things correct when cutting the
LP. e.g. use the right shape of cutter, heat the cutter, apply the correct
pre-emphasis, etc, etc.

So you can make a CD with no dither (or equivalent). You can also make an
LP without doing what is required for good results. In each case the
failure to do the process correctly may have an adverse impact on the
results.

From information theory, dither (or an equivalent) is *required* if you
wish the result of an analogue source being recorded to reproduce in a way
that is formally indistinguishiable from being analog. i.e The result is
then such that it shows no signs of ever having been in digital/sampled
form.

However: In every case - CD, LP, cassette tape, etc - what you get will
depend on how well each stage of the process was carried out. But 'digital'
is no different to any other format in this respect.

, objective and ubiquitous component of digital
recording then?

You would have to ask the people who create each CD (or LP) that. I have a
number of lousy or faulty CDs. However, by comparison, a far higher number
of the LPs I've had showed problems of various kinds.

And you don't have to reduce
bit depth (not rate) to 16 - you can record that way in the first
place. It offers far more signal to noise ratio than any available
analogue format. 24 bits is available because it is possible, and it
does offer a S/N advantage over 16 bits - not for an audio recording,
of course,

Main advantage in audio is that it gives you headroom so you can make
recordings without having to fret so much about clipping. Rather ironic
given how willing those who then make pop/rock CDs then are to
systematically trash the results. Alas, the problem here is with the people
making the decisions, not with the systems they control.



Yes, I'd picked up that recording is sometimes done in higher than 16
bits. There must be an algorithm I'd have thought. What you say
contradicts aspects of the Wikipedia entires. Have you never felt
inclined to put these right? It only takes a couple of seconds. Look up
'dither' and 'quantization' and see what you think. If you'd prefer I'll
pick out a few examples.

Afraid I have always felt that trying to 'correct wikipedia' is about as
sensible as herding cats. Life is too short. :-)


OK, you hate the sound from records, and that's really an end to it. I
get a lot of enjoyment from records - sound and artefact.

I just bought a 4 CD Stiff back catalogue collection. Compressed and
sort of 'dolbyised', with dull treble. I played the vinyl version of one
of the tracks I have. I honestly fail to believe the vinyl sounds worse.

Blame the people who made the CDs. Not the CD Audio format.

if you look in this month's HFN you'll see some more work on compression
and clipping on CDs. Some examples have effective dynamic ranges sic of
about 1dB due to level compression.


Listening to Amy Winehouse (more her band actually) on the Glastonbury
thing shows what can be done under very difficult conditions - I thought
the sound was superb.

For classical music, a good bet - as ever - is to listen to R3. Also to buy
BBC Music Magazine. Their cover CDs are often really excellent, both as
recordings and as performances. But I don't know of anything similar for
other types of music, alas.

Slainte,

Jim

--
Change 'noise' to 'jcgl' if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

Dirty Digital [sic.]
 

Arny Krueger wrote:
"Rob" wrote in message


Just seemed possibly ironic that digital might have to be
'shaped' to sound decent.

And you think that LPs aren't shaped?


I do think they are. The point I was trying to establish is that there's
only one way of doing digital. Skim and ill-informed reading indicates
that there *could be* variables - dither, bit rates, and digital to
analogue conversion. The consensus appears to be that these variables
are either irrelevant to sound, or objectively and universally applied.

All of which doesn't make either necessarily sound more enjoyable to me.

I gave a TT to a friend a while ago. He has a couple of records and
didn't see the harm in making house space. He never saw the point of
records on the whole, particularly if a CD was available. He's trained
in physics and computer science (first class degree and so forth) - he
likes his empirical rationality. We hooked up the TT and put a record
on, and it sounded *engaging*. We haven't discussed it fully, but I'm
sure he enjoys what comes out the other end. In his mind, it shouldn't
sound any more enjoyable than a CD - less so, in fact. But it can. And
he's kept the TT and actually asked me for a few records that he has on CD.


You've obviously had a bad experience of vinyl

Almost 40 years when it was for many practical purposes, all we had to
listen to.

It's moved on a lot in last 40 years IMO.

If you study the professional literature in the field of recording, you will
find that there have been no significant innovations in almost 30 years.


I think Thorens moved things along in the 60s and early 70s, and Pink
Triangle set a benchmark that hasn't really been bettered significantly
to this day. Rega RB250 and AT MCs, both 80s I think, similar.

If you read the consumer audio ragazines, a naive person might think
otherwise.


Indeed. I don't buy them but have a leaf through if I'm waiting for a train.

Rob

Dirty Digital [sic.]
 

David Looser wrote:
"Don Pearce" wrote in message
et...

Bit depth - How many bits make up a sample. for CD it is 16. TV sound
(NICAM) uses 10. Telephones use 8.

I think I ought to add that NICAM is 14 bit, and telephones 12 bit, before
compressing.



Yup, I was thinking of explaining that NICAM is a sort of "best 10 of
14" over a 1msec time slot thing, but that is probably a step too far
right now.

d

Dirty Digital [sic.]
 

I know this is discussed a lot on this NG, but as simply an 'audio
enthusiast' who likes a decent sound, I do find this frustrating. I
can't seem to get out of the habit of going in to these music boutiques,
and ending up disappointed most of the time.

Listening to Amy Winehouse (more her band actually) on the Glastonbury
thing shows what can be done under very difficult conditions - I thought
the sound was superb.

Rob

You should get out more to a few live concerts to remind what they
should be like. More often than not you'll want to "twiddle something"
as it doesn't sound quite "right"....

--
Tony Sayer

Dirty Digital [sic.]
 

In article , Rob
wrote:


Changing from say 24 to 16 bitrate involves a process, involving one or
more steps, and calculation: an algorithm.

As I think as already been pointed out, neither '16' not '24' is a
'bitrate'.

Bitrate (per channel) in this context is the product of the sampling rate
and the bit *depth*. CD Audio has a bit depth of 16 - i.e. 16 bits per
sample.

Of course if either effective bit depth or information rate is your guide,
then in both cases you'd be better off with well made CD than well made LP.
The snag is that in each case you will be offerred the recordings those
involved can be bothered to produce. Which might waste the potential of
either medium.

If you say all current DACs available are stunning and incapable of
differentiation then I'm in no position to mount a scientific challenge.

So far as I know the following is the case for commercial DACs produced
during the last decade or more.

Despite magazine reviews repeatly devoting pages to difference in the
'sound' of decent DACs, none of those involved have shown they can
distinguish one from another *when they only have the sounds produced to
use for their decision*. Indeed, I think there have been a number of tests
where people have simply failed to be able to reliably tell which one they
are listening to when their only guide is the sound.

Most recently were the test done by a USA group of people where, under
normal listening conditions - people could show any ability to identify
'high bitrate' signals from CD Audio ones. This test was published in JAES.
Of course, it was then dismissed by those who produce reviews as it 'must
be wrong'... :-)

Slainte,

Jim








Rob

--
Change 'noise' to 'jcgl' if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

Dirty Digital [sic.]
 

In article , Don
Pearce
wrote:
Rob wrote:


Couple of the Wki things - 24 bit recording isn't always dithered, and
there are different types of dither.


It is certainly dithered, but that doesn't necessarily mean that a
specific noise signal has been provided for the purpose. Anything
noise-like will dither, and at 24 bits the inevitable input noise from
the analogue circuitry is already more than big enough for the job. It
is only when native noise levels are below the digital threshold that
you need to dither specifically. And of course when you are creating
music by synth.

Also if doing something like using 24 bit source material to create
something at a different bit depth or rate. e..g using 24/96kHz recordings
to make a CD. In general, the default assumption should be to employ dither
(or an equivalent) when processing the recording in the digital domain.
Although I have no idea if those in the business know this, or perhap even
in many cases have a clue what the equipment inside the boxes does. But
with luck, the people who made their kit had the necessary level of clue.
:-)

Slainte,

Jim

--
Change 'noise' to 'jcgl' if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

Dirty Digital [sic.]
 

In article , Rob
wrote:
Eeyore wrote:



- or at least algorithmic?
Uh ?
Changing from say 24 to 16 bitrate involves a process, involving one
or more steps, and calculation: an algorithm.

Not it just gives more resolution. More than you can practically use
in fact. You get 6dB of s/n ratio per bit -3dB off the total.


OK, I'll have to put hands up and say you've completely lost me. I would
have thought changing 24 bit to 16 bit involves calculation.

It does - although it depends what you would call 'calculation'.

Examples.

A) You could just lop off the bottom 8 bits (trunkate). This 'calculation'
is essentially just an integer division, discarding the part below the
lowest bit, or a shift 8 places preserving sign.

B) You could add noise (dither) to the 16-bit level to the 24 bit values
and then lop off the bottom 8 bits (dither and trunkate) as per (A).

C) You could take the lowest 8 bits you lop off and use it as a 'remainder'
to modify later trunkations. Making use of the info that (A) discards. This
is called 'noise shaping' and can be done in a wild variety of ways. But
the result is nominally to preserve details that otherwise would be lost.
However you also need to dither unless you are also upping the sample rate.

I'd be inclined to avoid the 'cheap and cheerful' (A) option as it may end
in tears. It will just re-introduce the distortions you'd have got if you'd
made the original recording at 16 bit with no dither. :-)

I'd expect well-done conversions from 24 bit source material down to 16 bit
CDA to use dither, and ideally noise shaping as well, This can have the
added benefit of causing the output noise to be displaced to very high
frequencies, so being less audible. However if the source material had
noise already well above -90dB then the differences will be largely
irrelevant as the noise will dither the results when you do (A). So with
noisy material (A) may well pass. This did mean, though, that at least the
lowest 8 bits of the 24 source were noise, not information, in the first
place.

Note that 'resolution' has more than one meaning. It can mean for an
individual value in isolation. But with a series of samples it can mean
what can be resolved as a *pattern* using the series of values. It is quite
easy with a digital system like CD to have the ability to resolve signal
patterns or details below 1 bit in amplitude. Indeed, a routine test is to
use dithered signals below the 1 bit level to check the resolution and
linearity of a player at these low levels. This relies on dither and/or
noise shaping, and would be impossible without them. One reason why they
are vital for correct performance.

The above is important for audio as we don't listen to individual and
isolated sample values. We respond to the patterns made by series of
samples. Thus the above means we can, in appropriate circumstances, hear
sounds below the 1 bit level if the recording is dithered/noise shaped
decently. This is one of the respects in which correctly dithered (or
equivalent) digital is the same as analog.

All I can say is that I think I can detect differences in real world
sources - computers for example.

You may be right. Alas, people make claims like this all the time... but
when tests are done to see if they can do as they claimed then - based only
on the sounds produced - they tend to show no sign of being able. However
the computer audio system you use may be poor, and so distinguishable for
that reason.

Slainte,

Jim

--
Change 'noise' to 'jcgl' if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

Dirty Digital [sic.]
 

Don Pearce wrote:

I was unhappy with much of the mixing and balance from Glastonbury - far too
much drums for the most part.

Modern damn newbie mixing style, along with making the vocals just that little
bit too quiet to hear peroperly..

Graham

Dirty Digital [sic.]
 

Rob wrote:

Eeyore wrote:
Rob wrote:
Eeyore wrote:
Rob wrote:

I'll obviously defer here, but isn't dither a type of
sound 'shaping', in that it changes the sound?
No. It's merely a tiny bit of noise at ~ -90dB. Coul;d be less with modern
24 bit converters. Most real world signals self-dither anyway.
OK fine, it doesn't change the sound. My source is Wikipedia I'd afraid.

And isn't reducing bitrates to 16 something of an art
You mean bit depth. 96dB ? Not enough ?

Have you any idea how many dB vinyl and analogue tape's capable of ? Nowhere
remotely NEAR that. Plus you had to use Dolby A to get multitrack tape to
deliver an acceptable level of noise.
Bits as in amount of information. More bits more information.

Uh ? Even 16 bit audio beats analogue tape and disc hands down.


- or at least algorithmic?
Uh ?
Changing from say 24 to 16 bitrate involves a process, involving one or
more steps, and calculation: an algorithm.

Not it just gives more resolution. More than you can practically use in fact. You
get 6dB of s/n ratio per bit -3dB off the total.


OK, I'll have to put hands up and say you've completely lost me. I would
have thought changing 24 bit to 16 bit involves calculation.

Depends what you mean by 'changing' ! In simplistic terms it lops off the bottom 8
bits.


Nothing I can think of is 'just done'. And I would have thought it (24-16) gave
less resolution. Why record at a higher bit rate to get lower resolution?!

You use 24 bit to get *headroom* for one there's less chance of overload due to the
massive dynamic range AND to be state of the art. You can 'normalise' the final mix to
16 bit so it still doesn't 'overload' (i.e run out of headroom - equivalent to clipping
) and then convert to 16 bit which maintains the best possible quality for CD
mastering.


And DACs?
DACs are STUNNING these days. The very early ones were a bit stinky but
they've been fine for a good 20 years or so now.

If you say all current DACs available are stunning and incapable of
differentiation then I'm in no position to mount a scientific challenge.

Example ... state of the art .....
24-Bit Resolution
Analog Performance:
Dynamic Range:
132 dB (9 V RMS, Mono)
129 dB (4.5 V RMS, Stereo)
127 dB (2 V RMS, Stereo)
THD+N: 0.0004%
Differential Current Output: 7.8 mA p-p
8â—Š Oversampling Digital Filter:
Stop-Band Attenuation: ñ130 dB
Pass-Band Ripple: ±0.00001 dB
Sampling Frequency: 10 kHz to 200 kHz

http://focus.ti.com/docs/prod/folder.../dsd1794a.html

At 1/4 the price ......

24-Bit Resolution
Analog Performance:
Dynamic Range: 123 dB
THD+N: 0.0005%
Differential Current Output: 4 mA p-p
8â—Š Oversampling Digital Filter:
Stop-Band Attenuation: ñ98 dB
Pass-Band Ripple: ±0.0002 dB
Sampling Frequency: 10 kHz to 200 kHz

http://focus.ti.com/docs/prod/folder...t/pcm1796.html

And that's a mere £1.85 in small production quantities.

Them's the facts, and to my untrained eye there would appear to be very
little 'real world' difference.

All I can say is that I think I can detect differences in real world
sources - computers for example.

The sound cards in computers are complete **** in comparison.

Graham

Dirty Digital [sic.]
 

tony sayer wrote:
I know this is discussed a lot on this NG, but as simply an 'audio
enthusiast' who likes a decent sound, I do find this frustrating. I
can't seem to get out of the habit of going in to these music boutiques,
and ending up disappointed most of the time.

Listening to Amy Winehouse (more her band actually) on the Glastonbury
thing shows what can be done under very difficult conditions - I thought
the sound was superb.

Rob

You should get out more to a few live concerts to remind what they
should be like. More often than not you'll want to "twiddle something"
as it doesn't sound quite "right"....


I've been to a few classical concerts, but I'm not a huge classical fan
so they're a bit wasted on me. I did go to a master-class quartet thingy
a year ago and that was fascinating - and the sound was amazing.

On pop/rock gigs, I just like live music. On reflection I've seen some
regrettable performances but I've never had less than a great time. It's
more the occasion I think. And the thought of recreating a live
performance in my living room is simply never going to happen. Not sure
i'd want it if I could. So, I just go for a sound that pleases.

Rob