PDA

View Full Version : MP3 quality poor due to low volume levels, notcompression.



Simon Turner
2003-12-31, 19:52
Jason.. the immediate goal is to avoid clipping (distortion), not to
"normalise" the whole collection.
I am not "normalising", as I understand this can irreversably change the MP3
files. Replay Gain does not "normalise". I have learnt this from the
Hydrogen Audio, MP3Gain and Replay Gain web sites.
The problem seems to be that when files have had clipping removed by MP3Gain
the volume level is reduced and is then lower than hardware devices like the
iPod and the Squeezebox are designed for. So the output thses devices then
provide is not sufficient to properly drive the input of many hifi
amplifiers. This results in a thin and weedy sound. And the only way to get
any volume is to crank up the volume on the hifi amp much more than one
would for the other peripheral devices (CD, Tuner etc).

Buying a more powerful amplifier will not change anything, just give me a
louder weedy sound.

I am just looking for a reason why, in my case, the iPod and the Squeezebox
do not seem to provide as high a level input to my hifi amplifier as all my
other (more standard) devices.

I am perplexed that there are not many others who have the same problem and
of course this worries me. I would be quite interested if the person who
said something like "I can't see how ripping through EAC could introduce
clipping" would run his ripped files through MP3Gain
(http://mp3gain.sourceforge.net) and read the programs help file.

Perhaps I ought to disregard MP3gain (Replay Gain) and all that is
recommended at Hydrogen Audio and just increase the gain on my files to
levels above clipping... but... although I cannot hear ANY distortion due to
clipping when MY MP3 files are played through my computer the distortion is
very obvious when the same files are played on my iPod.

My post is not meant as a criticism of the Slim devices at all. I am just
looking for help.

Simon
Brighton UK.




-----Original Message-----
From: discuss-bounces (AT) lists (DOT) slimdevices.com
[mailto:discuss-bounces (AT) lists (DOT) slimdevices.com]On Behalf Of Jason
Sent: 31 December 2003 23:32
To: 'Slim Devices Discussion'
Subject: [slim] MP3 quality poor due to low volume levels,
notcompression.


What is the whole goal here? I just rip/encode with EAC and LAME. I have a
few albums that have slightly lower volume than others but I don't see the
need to normalize my whole collection down to 89db just to prevent having to
adjust the volume knob once in a while.

So again, what is the goal here? If you are doing normalization there are
lots of additional factors that can be screwing things up for you.

If you are saying that the squeeze is not loud enough once everything is
normalized, welcome to the wonderful world of normalization, you can always
buy a more powerful amplifier.

-----Original Message-----
From: discuss-bounces (AT) lists (DOT) slimdevices.com
[mailto:discuss-bounces (AT) lists (DOT) slimdevices.com] On Behalf Of Jason Holtzapple
Sent: Wednesday, December 31, 2003 2:19 PM
To: Slim Devices Discussion
Subject: [slim] MP3 quality poor due to low volume levels,
notcompression.

--- Caleb Epstein <cae (AT) bklyn (DOT) org> wrote:
> On Wed, Dec 31, 2003 at 08:40:20PM -0000, Simon Turner wrote:
>
> > Hi Jason, Volume level is set to max. I use Constant Gain as it
> > enables me to be sure I have got the cd I am ripping to the highest
> > volume without icurring clipping in any individual track, whilst
> > still retaining the different volume levels between tracks. Usually
> > i find that just going one step down (i.e. -1.5db) gets rid of any
> > clipping introduced in the ripping process. I'm not too sure if
> > album gain does exactly this.
>
> How could ripping a CD introduce clipping? You're just
> extracting the WAV data off of the disc. If you have a decent
> drive or use the right software, its a perfect copy.

Probably not the ripping, but the encoding. I have plenty of
EAC/LAME --alt-preset standard encodes that are clipped
'out of the box' according to mp3gain.

> I expect album gain normalizes an entire disc as a single
> unit, so the highest peak of the entire album is at 0db. You
> seem to be normalizing on a per-track basis, so the highest
> peak of every track is at 0db.

It might help if Simon tells us what he's trying to
accomplish: a similar volume throughout his collection
or simply trying to reduce clipping.

--Jason

__________________________________
Do you Yahoo!?
Find out what made the Top Yahoo! Searches of 2003
http://search.yahoo.com/top2003

Craig Brannan
2003-12-31, 21:00
Still trying to picture what exactly a weedy sound is...

but in any event - it seems like you're going through a somewhat
laborious process to eliminate distortion - yet that very process seems
to be adding it's own greater distortion (i.e. deviation from the
original source).

The easiest answer would be to abandon the extra steps. MP3 is a lossy
compression process and will therefore add some form of distortion to
your music. Is it possible that what you describe as clipping is just
the distortion (loss) inherent in the MP3? If so then no amount of gain
adjustment will correct it and the only real solution would be to
switch to a lossless encoding format.

Does everything from the squeeze sound weedy - or is it just the gain
adjusted files?

craig.

On Dec 31, 2003, at 9:52 PM, Simon Turner wrote:

> Jason.. the immediate goal is to avoid clipping (distortion), not to
> "normalise" the whole collection.
> I am not "normalising", as I understand this can irreversably change
> the MP3
> files. Replay Gain does not "normalise". I have learnt this from the
> Hydrogen Audio, MP3Gain and Replay Gain web sites.
> The problem seems to be that when files have had clipping removed by
> MP3Gain
> the volume level is reduced and is then lower than hardware devices
> like the
> iPod and the Squeezebox are designed for. So the output thses devices
> then
> provide is not sufficient to properly drive the input of many hifi
> amplifiers. This results in a thin and weedy sound. And the only way
> to get
> any volume is to crank up the volume on the hifi amp much more than one
> would for the other peripheral devices (CD, Tuner etc).
>
> Buying a more powerful amplifier will not change anything, just give
> me a
> louder weedy sound.
>
> I am just looking for a reason why, in my case, the iPod and the
> Squeezebox
> do not seem to provide as high a level input to my hifi amplifier as
> all my
> other (more standard) devices.
>
> I am perplexed that there are not many others who have the same
> problem and
> of course this worries me. I would be quite interested if the person
> who
> said something like "I can't see how ripping through EAC could
> introduce
> clipping" would run his ripped files through MP3Gain
> (http://mp3gain.sourceforge.net) and read the programs help file.
>
> Perhaps I ought to disregard MP3gain (Replay Gain) and all that is
> recommended at Hydrogen Audio and just increase the gain on my files to
> levels above clipping... but... although I cannot hear ANY distortion
> due to
> clipping when MY MP3 files are played through my computer the
> distortion is
> very obvious when the same files are played on my iPod.
>
> My post is not meant as a criticism of the Slim devices at all. I am
> just
> looking for help.
>
> Simon
> Brighton UK.
>
>
>
>
> -----Original Message-----
> From: discuss-bounces (AT) lists (DOT) slimdevices.com
> [mailto:discuss-bounces (AT) lists (DOT) slimdevices.com]On Behalf Of Jason
> Sent: 31 December 2003 23:32
> To: 'Slim Devices Discussion'
> Subject: [slim] MP3 quality poor due to low volume levels,
> notcompression.
>
>
> What is the whole goal here? I just rip/encode with EAC and LAME. I
> have a
> few albums that have slightly lower volume than others but I don't see
> the
> need to normalize my whole collection down to 89db just to prevent
> having to
> adjust the volume knob once in a while.
>
> So again, what is the goal here? If you are doing normalization there
> are
> lots of additional factors that can be screwing things up for you.
>
> If you are saying that the squeeze is not loud enough once everything
> is
> normalized, welcome to the wonderful world of normalization, you can
> always
> buy a more powerful amplifier.
>
> -----Original Message-----
> From: discuss-bounces (AT) lists (DOT) slimdevices.com
> [mailto:discuss-bounces (AT) lists (DOT) slimdevices.com] On Behalf Of Jason
> Holtzapple
> Sent: Wednesday, December 31, 2003 2:19 PM
> To: Slim Devices Discussion
> Subject: [slim] MP3 quality poor due to low volume levels,
> notcompression.
>
> --- Caleb Epstein <cae (AT) bklyn (DOT) org> wrote:
>> On Wed, Dec 31, 2003 at 08:40:20PM -0000, Simon Turner wrote:
>>
>>> Hi Jason, Volume level is set to max. I use Constant Gain as it
>>> enables me to be sure I have got the cd I am ripping to the highest
>>> volume without icurring clipping in any individual track, whilst
>>> still retaining the different volume levels between tracks. Usually
>>> i find that just going one step down (i.e. -1.5db) gets rid of any
>>> clipping introduced in the ripping process. I'm not too sure if
>>> album gain does exactly this.
>>
>> How could ripping a CD introduce clipping? You're just
>> extracting the WAV data off of the disc. If you have a decent
>> drive or use the right software, its a perfect copy.
>
> Probably not the ripping, but the encoding. I have plenty of
> EAC/LAME --alt-preset standard encodes that are clipped
> 'out of the box' according to mp3gain.
>
>> I expect album gain normalizes an entire disc as a single
>> unit, so the highest peak of the entire album is at 0db. You
>> seem to be normalizing on a per-track basis, so the highest
>> peak of every track is at 0db.
>
> It might help if Simon tells us what he's trying to
> accomplish: a similar volume throughout his collection
> or simply trying to reduce clipping.
>
> --Jason
>
> __________________________________
> Do you Yahoo!?
> Find out what made the Top Yahoo! Searches of 2003
> http://search.yahoo.com/top2003
>

roy@rant-central.com
2003-12-31, 23:03
On Wednesday 31 December 2003 11:00 pm, Craig Brannan wrote:
> Still trying to picture what exactly a weedy sound is...
>
> but in any event - it seems like you're going through a somewhat
> laborious process to eliminate distortion - yet that very process seems
> to be adding it's own greater distortion (i.e. deviation from the
> original source).
>
> The easiest answer would be to abandon the extra steps. MP3 is a lossy
> compression process and will therefore add some form of distortion to
> your music. Is it possible that what you describe as clipping is just
> the distortion (loss) inherent in the MP3? If so then no amount of gain
> adjustment will correct it and the only real solution would be to
> switch to a lossless encoding format.
>
> Does everything from the squeeze sound weedy - or is it just the gain
> adjusted files?

FWIW, I just completed an mp3gain pass over my whole ripped collection because
I was tired of reaching for the volume control. I didn't try to maximize
anything, but just processed everything an album at a time (to retain the
relative gain across each album's tracks). Most of my stuff was reduced.
(Barenaked Ladies' 'Stunt' dropped 5 dB) Some was boosted (Planet P Project
raised 1 dB) But all of it sounds as good as before processing. Only the
relative volume levels changed; there was no noticable change to the
character of the sound. And the relative output level is still comparable to
other inputs on my receiver.

Granted, my ears are far from golden. But considering that mp3gain does
nothing more than change the relative amplitude of the signal, I honestly
can't understand how that would change the percieved "quality" of the
reproduction. If Simon is truly experiencing a noticable change in timbre
such as he describes, I'd think that there is something besides gain
adjustment at play.

Nicolas Guillemain
2004-01-01, 08:21
Le 01/01/2004 à 06:35:17, Jason écrivait :

> MP3gain does not FIX clipping [...]

Well, it does - in a way.

Clipping is somehow inherent to MP3 decoding. As I understand it (but
don't quote me on it and check out hydrogenaudio.org), rounding errors
when the file is decoded are so that a source track that peaks at -0.1dB
will most probably clip at 0dB when decoded. It tends to be a problem
with most modern music, which is generally mastered with very limited
headroom, if any - so the resulting MP3s are very likely to clip when
played back, even if there's no actual clipping in any of the MP3 frames.

So you're right, using MP3Gain on these files does not actually "fix"
clipping, as there is no actual clipping to begin with, at least not in
the files themselves - what it does is apply ReplayGain normalization
(or "loudness" normalization, as opposed to peak normalization) in such
a way that playback will no longer induce clipping. It's a lossless and
completely reversible process, so you can always go back and adjust the
target volume if the result sounds too "thin" for you, whatever that
means.

As several posters already pointed out, hydrogenaudio.org is pretty much
the definitive reference on those tricky subjects. There's more info
there than you can shake a stick at - it's really worth checking out the
forums if you want to learn more on the fine art of audio encoding.

Also, the MP3Gain help file has a good primer on why loudness
normalization is better than standard peak normalization, again worth
checking out if you're not sure what ReplayGain utils can do for you.

Be seeing you,
Nicolas
--
http://www.grenouille.com - la météo du net

T
2004-01-01, 11:45
> I am perplexed that there are not many others who have the same problem
and
> of course this worries me. I would be quite interested if the person who
> said something like "I can't see how ripping through EAC could introduce
> clipping" would run his ripped files through MP3Gain
> (http://mp3gain.sourceforge.net) and read the programs help file.

Interesting, I just ran some 0dBFS mp3 files through mp3gain (various
sine-waves and sweeps, 20 files). It indicated that all of the files were
clipped, despite the fact that none of these files are clipped when playyed!
I design professional audio broadcast equipment for a living, so I do know
what I am talking about.

Tom