How does Freeview radio compare with DAB/FM?

In article , Dave Plowman (News)
wrote:
In article , DAB sounds worse than FM
wrote:
I've not listened to or checked BBC FM or DAB in some time. But on
previous occasions when I did DTTV had a measurably wider dynamic
range for Radio 3 than FM. I'd say this was also an audible
difference at times - when I listened seriously to R3 on FM. Now I
mainly use DTTV and iPlayer.


You go on and on about dynamic range compression, yet you ignore the
amount of quantisation noise that's added to the signal by encoding to
192 kbps MP2.

Get a grip.

One day you'll realise that people hear things differently - and what
annoys most between different types of distortion is personal.

You obviously like the sound of multipath.

It is curious that FM and LP share a tendency for the distortion to rise
with frequency and be worse for the L-R component. Maybe people like this,
or at least find it harder to hear. No idea, but a curious coincidence.

Slainte,

Jim

--
Please use the address on the audiomisc page 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

Jim Lesurf wrote:
In article , DAB sounds worse than FM
wrote:
Jim Lesurf wrote:
In article , Steve Thackery
wrote:

"DAB sounds worse than FM" wrote in message
...

FM with a good signal is better - it's far, far more accurate at
reproduction than audio on DAB/DTT.

Could we have some objective evidence for that? Frequency response,
harmonic distortion, intermod distortion, phase distortion, noise,
all the usual?

Whilst Steve is preparing his own detailed explanation of what he
meant, new readers might like to look at

http://www.audiomisc.co.uk/HFN/BandwidthBlues/page.html

which illustrates how using a finite transmission (or RX IF bandwith)
can cause distortion for FM stereo, Note that it ignores effects like
multipath and RX imperfections which can also lead to HF distortion in
practice.


You go on about distortion, Jim. How about the distortion introduced by
MP2 audio encoding??

I thought you had said in another posting that defining 'distortion' for
lossy encoding isn't a simple matter. So you'd first have to define how I
would measure or compute that for recorded music.


Mean squared error is a common way to quantify distortion in DSP. (Sampled)
FM would obviously beat DAB on that basis - think Parseval's theorem, i.e.
750 Hz-wide chunks of the spectrum are just discarded for DAB but not for
FM, and the bit rate reduction will also add massive amounts of error for
DAB but not for FM.

Then again, the number of times that 750 Hz-wide chunks are discarded for
lower frequencies isn't high, because with the subbands being so wide the
psychoacoustic model can't justify discarding lower frequency subbands very
often, which just goes to show how ridiculously inefficient the MP2 codec
is, because there's so little opportunity to discard sections of spectra
that transform codecs could discard.


That said, if you look at the reply I just made, I do give percentage
values for the bit-value differences / source for some mp3 examples. This
in some ways mimics expressing conventional distortion in percentage
terms. But can be quite misleading as simply finding differences on an
encode-decode cycle may be due to various other things like phase
alterations changing the waveform shape, or other factors. Hence just
displaying difference levels on a sample-by-sample basis doesn't really
tell us how audible the changes may be, or if the are 'distortion' in any
meaninful sense equivalent to the conventional use of that term.

Also, you never seem to stray away from Radio 3, yet 98% of stereo music
stations on DAB in the UK use a bit rate of 128 or now even 112 kbps.

I think I've mentioned it in the past, but you still seem not to have
noticed that for many, many months now I've not really listened to DAB[1].
Or apparently even my comments more recently that I normally listen to
DTTV and iPlayer, not DAB or FM.

Ultimately, if you can't hear the problems with the audio quality on
DAB, I fear that your hearing is screwed.

Maybe your problem is that you haven't yet realised that I stopped taking
any interest in listening to DAB ages ago. Happy to leave flogging horses
to you as you are so eager to spend your time on that. :-)


DAB or DTTV for Radio 3, in both cases you're trying to compare FM with 192
kbps MP2. I say again, you write all this stuff about distortion on FM, then
completely ignore the massive amount of quantisation noise that's added to
the signal with 192 kbps MP2 - the most inefficient widely-used audio codec.

It's comparing apples with ugly fruit, basically.

But you still think your stuff about distortion on FM means something.

Yeah, whatever you say, Jim.



--
Steve - www.digitalradiotech.co.uk - digital radio news & info

The BBC's "justification" of digital radio switchover is based on lies

Jim Lesurf wrote:
In article , DAB sounds worse than FM
wrote:
Steve Thackery wrote:
"DAB sounds worse than FM" wrote in message
...

FM with a good signal is better - it's far, far more accurate at
reproduction than audio on DAB/DTT.

Could we have some objective evidence for that?


Here's a plot of quantisation noise added to an audio signal by
compressing it to 192 kbps MP2:

http://www.digitalradiotech.co.uk/im...V_original.gif

Quantisation noise = encoded audio - uncompressed audio

(subtraction is performed on a sample by sample basis)

You don't really specify some of the details I'd find useful to assess the
above. I assume the vertical scale is the difference in sample values,


I did say that when I said quantisation noise = encoded audio - uncompressed
audio on a sample by sample basis.


but
you don't give a plot of the actual sizes of the original source samples.
Hence I can't tell how far down in relative terms (dB or percentage) what
you plot would be.


Does it really matter, Jim? Try to think about this logically. Reducing the
number of bits per sample from 16 to an average of just over 2 is pretty
much always going to produce massive amounts of quantisation noise if you're
encoding a complex waveform.


FWIW as a comparison people might like to look at

http://jcgl.orpheusweb.co.uk/temp/mp3err.gif

This is taken from an article that is just appearing (not in HFN) and
shows some results for 'mp3' encoding rather than 'mp2'. The encoding and
decoding were done with 'lame' with settings that seemed to give the best
results in terms of minimising the difference between the source and a
version passed though an encode-decode cycle. The values are given as
percentages. This is of amplitudes of the sample values, but taking the
ratio of the difference to the source before encoding. So more like a
'percentage' way of expressing conventional distortions. The plot shows
the peak size of the percentage during each 100ms of the data. The source
is orchestral music (from a prom concert) that typically peaks at about
-10 to -15dBFS during the first part of the section, falling to more like
-30dBFS during the latter part.

So if we were to interpret such sample differences caused by
encoding-decoding in a parallel way to conventional distortion mp3 would
not be impressive even at 320kbps! However this is likely to be a
misleading method as the waveform alterations on a sample-by-sample basis
aren't simple distortion in the traditional sense.


Yeah, whatever you say, Jim.


--
Steve - www.digitalradiotech.co.uk - digital radio news & info

The BBC's "justification" of digital radio switchover is based on lies

In article , DAB sounds worse than FM
wrote:
Jim Lesurf wrote:
In article , DAB sounds worse than FM
wrote:



You go on about distortion, Jim. How about the distortion introduced
by MP2 audio encoding??

I thought you had said in another posting that defining 'distortion'
for lossy encoding isn't a simple matter. So you'd first have to
define how I would measure or compute that for recorded music.


Mean squared error is a common way to quantify distortion in DSP.

Alas the problem with that is almost any changes to the waveform can look
like 'distortion' even if they are entirely linear. So I can't say that
seems a good choice to me for music though the systems we are discussing.
Indeed, from the plots I just put up 320kbps mp3 would also seem very poor
if we assumed the sample differences were 'distortion' in the same sense as
traditionally used.

[snip]


Maybe your problem is that you haven't yet realised that I stopped
taking any interest in listening to DAB ages ago. Happy to leave
flogging horses to you as you are so eager to spend your time on that.
:-)


DAB or DTTV for Radio 3, in both cases you're trying to compare FM with
192 kbps MP2.

The comparisons I referenced here earlier were measurements on DTTV versus
iPlayer. Not FM.

Seperately I pointed out an example of the distortion mechanisms that
affect FM.

Lost count of how many times I've said I don't take much interest these
days in either FM or DAB.

I say again, you write all this stuff about distortion on
FM, then completely ignore the massive amount of quantisation noise
that's added to the signal with 192 kbps MP2 - the most inefficient
widely-used audio codec.

You haven't actually shown that the differences in your graph *are* either
"distortion" or "quantisation noise". Nor - so far as I have seen - have
you yet responded to the comments I made re points like your not giving any
info on the relative sizes of the original values with which we'd need to
compare the plotted 'differences' you graphed to establish how large they
were relative to the music. It is usual in audio to specify 'distortion'
in relative units like percentage or dB for fairly obvious reasons. It is
also usual to exclude things like linear changes from being 'distortion'.

Slainte,

Jim

--
Please use the address on the audiomisc page 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

"Hugh Newbury" wrote in message
...
Subject says it all. I get radio on my tv, and it sounds OK. But how does
it really compare? And does it make any difference putting the output into
an audio system?

Hugh (confused)

--

Hugh Newbury

www.evershot-weather.org

Hello Hugh.

Radio 7 is in stereo on Freeview and in mono on DAB.

Regards, Roger.



--- news://freenews.netfront.net/ - complaints: ---

Hang on Jim, isn't FM radio supposed to be limited to 15KHz top
audio response anyway? ISTR that's why Auntie used the Goldring
G800 cartridges as they were robust, inexpensive, and faded out
quite quickly above 15K.

Robust would be the most important consideration as remember the worst
type of distortion is groove jumping;!...


Either way the audio has to be filtered out so as not to impose
upon the 19KHz pilot tone. Allow the audio to go above 15K and
then insert a very steep filter and it will have phase effects
well down into the audible band.




--
Tony Sayer

Almost certainly down to the processing being rather differently set..

Well in my case the main reason I largely gave up FM in favour of DTTV and
iPlayer was a combination of two factors.

Level compression. When I first started comparing FM with DTTV (and at that
point DAB) I initially preferred FM as it sounded 'nicer'. However over a
few months I listened to all three types of broadcast. Over that time I
started to realise that what I was noticing most was that FM radio 3 was
move level compressed than DTTV/DAB. Once I became more familiar with it I
preferred the wider dynamic range of DTTV. The process was interesting as I
realised I'd previously become 'habituated' to the compression on Radio 3
FM and had come to 'expect music to sound like that'.

Interference and link problems. I have always found that FM tends to suffer
from background interference and other effects like multipath.

Well I often wonder of the feed to your local FM TX is as it should be
or your aerial system is what it needs be..

The
interference can be quite distracting for serious radio 3 listening as the
sound level is low for much of the time. This also means that - even with a
good signal and RX - background noise was noticable on FM when it wasn't
on DTTV.

Must be a duff feed if its that noticeable..

Since then I've come to like the iPlayer simply because I can listen as and
when I want.

So I'd say it was 50:50 on your processing point. Yes, I think I noticed
the extra compression on FM, although I can see it was needed to avoid
quiet passages sinking into noise. But it was also effects like
interference and HF distortions.


Do any of the artefacts on the BBC digital annoy you at all?..


--
Tony Sayer

In article , Mark Carver
scribeth thus
DAB sounds worse than FM wrote:

Yes, BBC FM is brickwall filtered at about 15.5 kHz IIRC.

The IBA COP always specified a brickwall at 15.00 kHz, I don't think the BBC
or any other EBU member spec differed ?

Normally taken care of in the processor in most all instances..



--
Tony Sayer

In article , Dave Plowman (News)
scribeth thus
In article ,
DAB sounds worse than FM wrote:
I've not listened to or checked BBC FM or DAB in some time. But on
previous occasions when I did DTTV had a measurably wider dynamic
range for Radio 3 than FM. I'd say this was also an audible difference
at times - when I listened seriously to R3 on FM. Now I mainly use
DTTV and iPlayer.


You go on and on about dynamic range compression, yet you ignore the
amount of quantisation noise that's added to the signal by encoding to
192 kbps MP2.

Get a grip.

One day you'll realise that people hear things differently - and what
annoys most between different types of distortion is personal.

You obviously like the sound of multipath.

Yes Dave but the whole county isn't riddled with multipath;!..
--
Tony Sayer

"DAB sounds worse than FM" wrote in message
...

Does it really matter, Jim? Try to think about this logically. Reducing
the number of bits per sample from 16 to an average of just over 2 is
pretty much always going to produce massive amounts of quantisation noise
if you're encoding a complex waveform.

Something doesn't sound right here. It goes without saying that *encoding*
(i.e. analogue to digital conversion of) an individual analogue datum into a
16-bit sample will produce much less quantisation noise that encoding it
into a 2-bit sample. The latter would be absurd, producing a digital
representation indistinguishable from a square wave.

Surely you are talking about an average bit rate of just over twice the
original sample rate AFTER COMPRESSION. Compression is completely different
from encoding, and the resultant artefacts are completely different, too.

So no, you can't possibly say that compressing an audio stream until the bit
rate is twice the sample rate is in ANY way analogous to reducing 16-bit
quantisation into a 2-bit quantisation.

Apples and oranges.

Here is a logical extension to the compression story: it is often easier to
think of the compressed audio stream NOT as compressed audio per se, but as
a series of instructions on how the receiver should reproduce the original
sound. For instance, if you were prepared to lose all of the original
nuances of the music, you could encode it on the fly as a MIDI data stream
(a great example of a series of instructions to the receiver, rather than
anything resembling the original sound waveform). The music would be
reproduced at the far end, but would be very different indeed from the
original. However, the bit rate would be extremely low indeed - a few Kb/s.

Furthermore, it would display no quantisation noise at all.

I'm using this as an extreme example to illustrate the fact that reducing
the bit rate using some sort of compression algorithm does NOT, of itself,
have anything to do with quantisation noise.

SteveT