Hi,

I've just put up a new page

http://www.audiomisc.co.uk/Linux/Sou...stenAgain.html

This discusses how I setup a (Ubuntu) Linux box to access the BBC iPlayer,
and provides some measurements on the results that look at some aspects of
the BBC aac/aac+ radio iPlayer.

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:
Hi,

I've just put up a new page

http://www.audiomisc.co.uk/Linux/Sou...stenAgain.html

This discusses how I setup a (Ubuntu) Linux box to access the BBC iPlayer,
and provides some measurements on the results that look at some aspects of
the BBC aac/aac+ radio iPlayer.


FWIW I've just built a xubuntu box to run on an old Celeron 500MHz CPU
(spawn of Pentium II), 512MB - It's a Compaq SFF PC from the skip. The
onboard sound was pants, so found an old Aureal Vortex 2 soundcard that
had no hope of further software support in Windows and used that to feed
the workshop amplifier.

I'm running Wine and Spotify on top. It sounds excellent. What's radio?

:-)

--
Adrian C

On Mon, 20 Jul 2009 15:13:56 +0100, Jim Lesurf
wrote:

Hi,

I've just put up a new page

http://www.audiomisc.co.uk/Linux/Sou...stenAgain.html

This discusses how I setup a (Ubuntu) Linux box to access the BBC iPlayer,
and provides some measurements on the results that look at some aspects of
the BBC aac/aac+ radio iPlayer.

Slainte,

Jim

The line thickening (and LF zigzag) - is that an artefact of the FFT,
or is threre really some modulation going on?

Your warnings about the distortion in the headphone output, but I'm
guessing that most people will use the line output, which I presume is
clean.

d

In article 4a6881f4.21074781@localhost, Don Pearce
wrote:
On Mon, 20 Jul 2009 15:13:56 +0100, Jim Lesurf
wrote:

Hi,

I've just put up a new page

http://www.audiomisc.co.uk/Linux/Sou...stenAgain.html

This discusses how I setup a (Ubuntu) Linux box to access the BBC
iPlayer, and provides some measurements on the results that look at
some aspects of the BBC aac/aac+ radio iPlayer.

Slainte,

Jim

The line thickening (and LF zigzag) - is that an artefact of the FFT, or
is threre really some modulation going on?

I think that is an artifact of my using simple 'triangular' apodisation
windowing. Plus, I think the finite resolution of the data. (I use a series
of FFTs power averaged to improve the SNR.

Fortunately the test tone stream the BBC/Siemens provided was dithered even
if the Flash plugin doesn't apply dither when computing volume level
rescaling.)

The line may appear thicker in some places as a result of the graphics
anti-aliasing where the angle or some other detail changes. That doesn't
show up with the vector graphic source. (All my graphics are generated as
'drawfiles' which are like a subset of PostScript line vector/object
commands. But I then have to generate a bitmap for the website versions.
That always degrades the appearance a bit. )

Your warnings about the distortion in the headphone output, but I'm
guessing that most people will use the line output, which I presume is
clean.

I don't know. But my experience makes me reluctant to presume line output
will be clean. So I'd say "maybe" to that.

The *only* analogue out for the laptop I used is the headphone. So I'd have
expected that not to clip when fed 0dB. In fact it clips at a far lower
level than 0dB. This is when it wasn't even driving a load with an
impedance anything like as low as a typical headset.

When I listened with a pair of headphones the sound was very loud unless
you wound down the volume a lot. So the gain applied between the internal
DAC and the headphone output stage was needlessly high. Would have been
fine if this had been reduced so 0dB didn't clip. So it all seemed like
careless design to me. The usual idiotic motto: "louder is better".

So for all I know some computer sound systems will also clip the 'line'
output. I can just warn people to be wary just in case. Check if you can.
Arny may know more about this. I've only been checking computer systems
recently once I'd developed and interest.

FWIW if I can blag a borrow I will try out some other USB 'soundcards' to
see how they work (if at all!) with Linux. Then report on what I find out.
A number of these have appeared recently. But I've only seen limited
results based on windows and macs.

The DACMagic is excellent in my opinion. I would recommend it to anyone who
has doubts about their existing soundcard. Subject to the caveat that I
can't say of someone else's hardware or OS version might not be willing to
co-operate. I had to change to Pulse Audio to get the iPlayer Flash plugin
to use it. So there may be other snags I've not encountered. This is the
problem with computer based systems. There are all too many hidden snags.

I've been quite surprised by how useless most magazine 'reviews' of
computers are in these respects. The seem to never do any measurements on
the output. So never say if the correct sampling rate is used, or if there
are any needless conversion artifacts, or even if the machine is
mechanically silent. About all you tend to get in the magazines I've seen
is a general 'sound works' comment - which means little more than they
could hear something! That may have been OK years ago. But now people are
increasingly using computers as sound sources for audio this should be
taken far more seriously IMHO.

Otherwise we may have people deciding that computer audio, and things like
the iPlayer have inherently poor sound, when the reality is that their
setup is mangling the sound. The solution in magazines is to spend
thousands on dedicated special purpose systems. But my own experiments
show you can get good results without spending vast sums. Alas, there is a
lack of the information needed to identify the correct kit and how to set
it up.

For example. I've been looking for some time now for a small laptop that
fits the bill. The requirements seem simple to me;

1) Must be mechanically silent. No fans or hard disc noises.

2) Fairly small so can be used like a radio tuner, etc.

3) Digital (ideally optical) output that gives the correct sample rate for
the source material, with no needless 'reconversions'. Thus able to output
bit-perfect results for LPCM sources.

4) Works with Linux. Ideally Ubuntu family as that is the popular choice at
the moment.

You'd think makers would currently be falling over themselves in the rush
to offer such systems. Nope. Not that I can see. Only a few ultra-costly
machines with specialist setups.

You'd think magazine reviews would be identifying them. Nope.

I was tempted by the fit-pc2 as it is said to run with Linux, is small and
silent, and I can use an external DAC. But I've since been told it doesn't
currently work with Ubuntu 9.04 and has various hardware snags. So no cigar
unless the snags are sorted.

Ditto, I wondered about the Dell Mini 9 as that got a decent review in a
linux mag. But I've since been told it also has snags, and has been
replaced as a model!

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

In article , Adrian C
wrote:
Jim Lesurf wrote:
Hi,

I've just put up a new page

http://www.audiomisc.co.uk/Linux/Sou...stenAgain.html

This discusses how I setup a (Ubuntu) Linux box to access the BBC
iPlayer, and provides some measurements on the results that look at
some aspects of the BBC aac/aac+ radio iPlayer.


FWIW I've just built a xubuntu box to run on an old Celeron 500MHz CPU
(spawn of Pentium II), 512MB - It's a Compaq SFF PC from the skip. The
onboard sound was pants, so found an old Aureal Vortex 2 soundcard that
had no hope of further software support in Windows and used that to feed
the workshop amplifier.

I should admit that I do prefer Xubuntu, but stayed with Ubuntu on the
Shuttle I'm using. Xfce is more like the managers of yore that I used
mumble years ago. Gnome is a bit too 'eye candy' for my taste.

Rather than switch to Xubuntu I am slowly displacing Ubuntu's gnome desktop
by ROX. :-) If you've not tried that, have a look. It is available via
synaptic as well as the other usual methods.

This should mean that eventually I have access to all the usual Ubuntu
apps, etc, but can have the ROX GUI/filer which is far more to my taste
than gnome's normal behaviour. Gnome is probably too much like windows for
me. :-) I prefer the simpler and more self-contained behaviour of ROX.
But this is probably because it is based on the way my preferred Risc OS
GUI functions. Helps me to feel at home with using a GUI on top of *nix.

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

On Mon, 20 Jul 2009 16:41:41 +0100, Jim Lesurf
wrote:

Your warnings about the distortion in the headphone output, but I'm
guessing that most people will use the line output, which I presume is
clean.

I don't know. But my experience makes me reluctant to presume line output
will be clean. So I'd say "maybe" to that.


Well, my card (which is on the decent side of medium) is an M-Audio
24/96, and seems perfectly happy flat out.

The *only* analogue out for the laptop I used is the headphone. So I'd have
expected that not to clip when fed 0dB. In fact it clips at a far lower
level than 0dB. This is when it wasn't even driving a load with an
impedance anything like as low as a typical headset.

When I listened with a pair of headphones the sound was very loud unless
you wound down the volume a lot. So the gain applied between the internal
DAC and the headphone output stage was needlessly high. Would have been
fine if this had been reduced so 0dB didn't clip. So it all seemed like
careless design to me. The usual idiotic motto: "louder is better".


None of that surprises me. Think of it like the volume control on any
amplifier. You don't expect the output to be just clipping from rated
input when the volume control is at max. If it did, most music would
be incapable of driving the amp to clipping. There is always some
slack, and your laptop's headphone socket shows that.

So for all I know some computer sound systems will also clip the 'line'
output. I can just warn people to be wary just in case. Check if you can.
Arny may know more about this. I've only been checking computer systems
recently once I'd developed and interest.

FWIW if I can blag a borrow I will try out some other USB 'soundcards' to
see how they work (if at all!) with Linux. Then report on what I find out.
A number of these have appeared recently. But I've only seen limited
results based on windows and macs.


Can you even get drivers for most of them? I use Asio here- I suppose
that is implemented in Linux as well?

d

In article 4a6993d7.25650265@localhost, Don Pearce
wrote:
On Mon, 20 Jul 2009 16:41:41 +0100, Jim Lesurf
wrote:

Your warnings about the distortion in the headphone output, but I'm
guessing that most people will use the line output, which I presume
is clean.

I don't know. But my experience makes me reluctant to presume line
output will be clean. So I'd say "maybe" to that.


Well, my card (which is on the decent side of medium) is an M-Audio
24/96, and seems perfectly happy flat out.

That may well be the usual situation. But I have no data so can't be sure.

When I listened with a pair of headphones the sound was very loud
unless you wound down the volume a lot. So the gain applied between the
internal DAC and the headphone output stage was needlessly high. Would
have been fine if this had been reduced so 0dB didn't clip. So it all
seemed like careless design to me. The usual idiotic motto: "louder is
better".


None of that surprises me. Think of it like the volume control on any
amplifier. You don't expect the output to be just clipping from rated
input when the volume control is at max. If it did, most music would be
incapable of driving the amp to clipping. There is always some slack,
and your laptop's headphone socket shows that.

The distinction is that in this case the same 'volume control' adjusts both
the analog and digital outputs. And if you set the gain to '0dB' to be able
to obtain bit-perfect LPCM digital output with no conversions then the
headphone output will clip. Turn down the volume to avoid clipping the
headphones, and you can't get 0dB output from the digital output, and all
the values are scaled down needlessly

That seems crazy to me.


FWIW if I can blag a borrow I will try out some other USB 'soundcards'
to see how they work (if at all!) with Linux. Then report on what I
find out. A number of these have appeared recently. But I've only seen
limited results based on windows and macs.


Can you even get drivers for most of them? I use Asio here- I suppose
that is implemented in Linux as well?

Pass. Drivers can be a problem. Fortunately, a number of people who use
Linus spend time trying to reverse-engineer the details that some makers
won't release.

There are lists of devices for which drivers are available. The snag is as
I have mentioned, though. e.g. 'It works' does not guarantee that the sound
device can output bit perfect at the correct sample rate. Just that 'you
can hear something'. But then you can probably say this for various
hardware under windows as well. The industry seems to assume few buyers
will notice or care and 'canna be bothered' I guess.

Personally, I think it should be a legal obligation for makers to either
provide drivers or the info needed to write them. Refusing to do so is an
anti-competitive practice in my view. As such I suspect it may be illegal
in the EU, and probably also in the USA. If not, then it should be IMHO.

But my impression is that most devices that are popular soon have someone
manage to write a driver.

FWIW Canonical (the people promoting Ubuntu) are keen on ensuring things
like drivers appear as they want Unbuntu Linux to be a choice that most
non-geeks can be happy to use. The current situation means that drivers for
audio and video are more important than in past decades, as is a decent GUI
for those who are deterred by terminals and arkane commands. My impression
is that they have made progress in these areas. But there are still
problems with some makers and devices.

FWIW Cambridge Audio had already checked, and were able to tell me their
DACmagic worked OK with Ubuntu before I tried one. It uses a standard USB
'headphone' chip for receiving audio via USB. I suspect other 'audiophile'
units do as it is the easiest way for companies with no computing
background to get into the game. I will find out if I can experiment in
future. But given the availability of the DACMagic if I need another I'd
happily choose one if nothing else I preferred looked like working. Direct
any cash to makers who ensure their equipment does the required task! :-)

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 4a6881f4.21074781@localhost, Don Pearce
wrote:
On Mon, 20 Jul 2009 15:13:56 +0100, Jim Lesurf
wrote:

Hi,

I've just put up a new page

http://www.audiomisc.co.uk/Linux/Sou...stenAgain.html

This discusses how I setup a (Ubuntu) Linux box to access the BBC
iPlayer, and provides some measurements on the results that look at
some aspects of the BBC aac/aac+ radio iPlayer.

Slainte,

Jim


Interesting - although I don't follow a lot of the technical text.

For reasons I don't need to understand, I take it it's better not to use
any volume control on the computer that affects the volume of a digital
output other than set it at maximum?

snip line thickening explanation, fascinating though it must be :-)


FWIW if I can blag a borrow I will try out some other USB 'soundcards' to
see how they work (if at all!) with Linux. Then report on what I find out.
A number of these have appeared recently. But I've only seen limited
results based on windows and macs.


If I get round to it, and you can return it at some point, you can
gladly borrow an Opcode Sonicport (bus powered USB DAC). Works natively
on a Mac and Windows Vista (but not XP). I got it from a recording place
round the back of Denmark St - I'd guess it had a sort of claim to
goodness at some point.


The DACMagic is excellent in my opinion. I would recommend it to anyone who
has doubts about their existing soundcard. Subject to the caveat that I
can't say of someone else's hardware or OS version might not be willing to
co-operate. I had to change to Pulse Audio to get the iPlayer Flash plugin
to use it. So there may be other snags I've not encountered. This is the
problem with computer based systems. There are all too many hidden snags.


Is it too much to ask for a standard? Anyhow, do you mean this:

http://www.cambridgeaudio.com/summar...&Title=Summary

It's £250. Ebay has DACs for £30 - Mini HIFI SPDIF+OPTICAL
DAC--PCM1793+DIR9001+OPA2134. What's the critical part of the component
here? It's not the DAC itself presumably - it's the analogue amplification?

I (still!) have problems understanding why one digital component
necessarily sounds different compared to another *unless* it's to do
with analogue amplification. And I don't really follow that - just seems
to make sense.


I've been quite surprised by how useless most magazine 'reviews' of
computers are in these respects. The seem to never do any measurements on
the output. So never say if the correct sampling rate is used, or if there
are any needless conversion artifacts, or even if the machine is
mechanically silent. About all you tend to get in the magazines I've seen
is a general 'sound works' comment - which means little more than they
could hear something! That may have been OK years ago. But now people are
increasingly using computers as sound sources for audio this should be
taken far more seriously IMHO.

Otherwise we may have people deciding that computer audio, and things like
the iPlayer have inherently poor sound, when the reality is that their
setup is mangling the sound. The solution in magazines is to spend
thousands on dedicated special purpose systems. But my own experiments
show you can get good results without spending vast sums. Alas, there is a
lack of the information needed to identify the correct kit and how to set
it up.


Agreed. Magazines seem to be little more than paper boutiques.
Commodities in general seem to have been 'Top Geared' - dumbed down to a
badge, spurious figures and presentation.

For example. I've been looking for some time now for a small laptop that
fits the bill. The requirements seem simple to me;

1) Must be mechanically silent. No fans or hard disc noises.

2) Fairly small so can be used like a radio tuner, etc.

3) Digital (ideally optical) output that gives the correct sample rate for
the source material, with no needless 'reconversions'. Thus able to output
bit-perfect results for LPCM sources.

4) Works with Linux. Ideally Ubuntu family as that is the popular choice at
the moment.

You'd think makers would currently be falling over themselves in the rush
to offer such systems. Nope. Not that I can see. Only a few ultra-costly
machines with specialist setups.


Apart from (4), me too. I tried for quite a while to get a quiet PC. A
nerdy friend has got close by using a 'brick' PSU, but there's no way
modern CPUs that can 'do video' will be cooled by convection alone. I
bought a net PC the other week - very impressive in lots of ways but
still needs a small fan.

You'd think magazine reviews would be identifying them. Nope.

I was tempted by the fit-pc2 as it is said to run with Linux, is small and
silent, and I can use an external DAC. But I've since been told it doesn't
currently work with Ubuntu 9.04 and has various hardware snags. So no cigar
unless the snags are sorted.


Ah - that's me wrong again! The netbook has an Atom processor and is
pretty clunky with Windows XP - and as I say, uses a fan which does kick
in frequently. Seems those Fit-PC2s have some sort of clever convection
cooling perhaps. Macs have very impressive cooling IME. Even the desktop
iMac (dual core Intel, fast HD, and decent graphics) has 3 fans that
'trickle' at around 1200rpm and produce a very unobtrusive sshhh. The HD
is silent. Room temp in here is 23C, CPU is 38C.

Rob

"Don Pearce" wrote in message
news:4a6993d7.25650265@localhost...
On Mon, 20 Jul 2009 16:41:41 +0100, Jim Lesurf
wrote:


None of that surprises me. Think of it like the volume control on any
amplifier. You don't expect the output to be just clipping from rated
input when the volume control is at max. If it did, most music would
be incapable of driving the amp to clipping. There is always some
slack, and your laptop's headphone socket shows that.


The rated input *is* the input which just drives the amp to clipping at
maximum volume. Whether "most music" does, or not, depends entirely on the
output level of the source. With RIAA inputs in general the rated input used
with typical MM cartidges means that even the very quietest records drive
the amp to clipping at maximum volume. With a typical pressing the amp will
be driven to clipping at some 20dB or so *below* maximum volume setting. I
remember an article in one of the HiFi mags many moons ago that concluded
that RIAA amps needed around 60dB overload margin based on "worst-case"
parameters. The fact that most HiFi amps of the time had around 20dB too
much gain didn't occur to the writer.

Similarly with inbuilt tuners which generally drive the amp to clipping at
around 30% modulation with maximum volume setting.

David.
..

In article , Rob
wrote:
Jim Lesurf wrote:
In article 4a6881f4.21074781@localhost, Don Pearce
wrote:


http://www.audiomisc.co.uk/Linux/Sou...stenAgain.html

This discusses how I setup a (Ubuntu) Linux box to access the BBC
iPlayer, and provides some measurements on the results that look at
some aspects of the BBC aac/aac+ radio iPlayer.


Interesting - although I don't follow a lot of the technical text.

For reasons I don't need to understand, I take it it's better not to use
any volume control on the computer that affects the volume of a digital
output other than set it at maximum?

Alas, not quite. The trick is to establish the settings for the computer
and its software that pass though digital values with no rescaling of the
amplitude of the values. This may be 'maximum', but that depends entirely
on how the programmers wrote the software, etc. If they don't tell you, you
are left to guess, or experiment, or measure.

With the specific systems I tried - that were Linux based - I found that
setting levels to '100%' (or '0dB') did this. So that is a good bet. But as
shown by the problem I found with headphones you can't be certain in
advance that some pinheaded software or hardware engineer hasn't upset this
and some other value is better.


FWIW if I can blag a borrow I will try out some other USB 'soundcards'
to see how they work (if at all!) with Linux. Then report on what I
find out. A number of these have appeared recently. But I've only seen
limited results based on windows and macs.


If I get round to it, and you can return it at some point, you can
gladly borrow an Opcode Sonicport (bus powered USB DAC). Works natively
on a Mac and Windows Vista (but not XP). I got it from a recording place
round the back of Denmark St - I'd guess it had a sort of claim to
goodness at some point.

I may be interested at some point. Afraid I've not heard of it. Is it a
currently available USB DAC? How does it identify itself on a windows box?


The DACMagic is excellent in my opinion.


[snip]

Is it too much to ask for a standard?

Afraid I don't know what you are asking here.

Anyhow, do you mean this:

http://www.cambridgeaudio.com/summar...&Title=Summary

Yes.

It's £250. Ebay has DACs for £30 - Mini HIFI SPDIF+OPTICAL
DAC--PCM1793+DIR9001+OPA2134. What's the critical part of the component
here? It's not the DAC itself presumably - it's the analogue
amplification?

No, the vital part in this context is the chip or other interface that
receives the USB audio. It has to identify itself to the computer in a way
that allows the data transfers to proceed correctly.

I (still!) have problems understanding why one digital component
necessarily sounds different compared to another *unless* it's to do
with analogue amplification. And I don't really follow that - just seems
to make sense.

The problem with USB DACs is as above. But with computer based systems all
kinds of other things can go wrong. For example the stream of values may be
'resampled' or otherwise fudged about by the system. So that what comes out
isn't what went in, even as a series of digital values.

I was tempted by the fit-pc2 as it is said to run with Linux, is small
and silent, and I can use an external DAC. But I've since been told it
doesn't currently work with Ubuntu 9.04 and has various hardware
snags. So no cigar unless the snags are sorted.


Ah - that's me wrong again! The netbook has an Atom processor and is
pretty clunky with Windows XP - and as I say, uses a fan which does kick
in frequently.

The good news is that many Linux distrbutions are somewhat less demanding
than Windows, so don't stress machines as much. But that may not matter
much if you then run tasks that are demanding. e.g. Processing video is
likely to be demanding regardless of OS. But I'd suspect that even a very
low spec machine should cope with simply passing around LPCM with no
processing without having to struggle.

Seems those Fit-PC2s have some sort of clever convection
cooling perhaps. Macs have very impressive cooling IME. Even the
desktop iMac (dual core Intel, fast HD, and decent graphics) has 3 fans
that 'trickle' at around 1200rpm and produce a very unobtrusive sshhh.
The HD is silent. Room temp in here is 23C, CPU is 38C.

The Shuttle I am using has a large, low speed fan that is meant to be
'quiet'. But I disconnected it to ensure silence. 8-] This is OK as what
I use the machine for isn't demanding for long periods, so only runs the
CPU at a small fraction of the level it can nominally deliver.

I also had a SSD fitted, not a traditional Magnetic HD. So no HD noises,
either.

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