[Mike Duffy]

Greetings, all.

I have a bunch of MP3 files I put on my phone (to play while in the car via
bluetooth.)

The problem is, they came from different sources, and some are louder than
others, and it's a PITA to keep changing the volume.

I found a neat set of GNU command line audio utilities (SOX). I wrote a
messy CMD command file to extract the RMS average decibel level of each
song, then used the formula:

10 ^ ((-20 - %RMSLEVEL%) / 20) (I wrote a 'quick&dirty' c++ for this.)

to get the volume (amplitude) multiplier for each song to bring it to -20,
which is pretty close to the aggregate average of the entire collection.
Then I ran an overnight script to adjust the volume of each MP3 file as
calculated above.

Then, just for fun, I re-ran the original script that calculates the RMS db
levels. I expected each song to be exactly -20. They are all pretty close
(around -20.5 db), so the volumes are close enough that I've solved my
problem. But I am curious. why are they not exactly at -20 db average?

There was some clipping for the songs that needed their volume increased.
Probably these were just scratches on the original vinyl. But even the ones
where volumes were lowered did not go exactly to -20.

Am I experiencing a compression-related syndrome? There are tools that let
you monkey with this as well, but I probably won't mess with that. I'm just
curious.

[geoff]

On 09/02/2016 17:34, Mike Duffy wrote:
Greetings, all.

I have a bunch of MP3 files I put on my phone (to play while in the car via
bluetooth.)

The problem is, they came from different sources, and some are louder than
others, and it's a PITA to keep changing the volume.

I found a neat set of GNU command line audio utilities (SOX). I wrote a
messy CMD command file to extract the RMS average decibel level of each
song, then used the formula:

10 ^ ((-20 - %RMSLEVEL%) / 20) (I wrote a 'quick&dirty' c++ for this.)

to get the volume (amplitude) multiplier for each song to bring it to -20,
which is pretty close to the aggregate average of the entire collection.
Then I ran an overnight script to adjust the volume of each MP3 file as
calculated above.

Then, just for fun, I re-ran the original script that calculates the RMS db
levels. I expected each song to be exactly -20. They are all pretty close
(around -20.5 db), so the volumes are close enough that I've solved my
problem. But I am curious. why are they not exactly at -20 db average?

There was some clipping for the songs that needed their volume increased.
Probably these were just scratches on the original vinyl. But even the ones
where volumes were lowered did not go exactly to -20.

Am I experiencing a compression-related syndrome? There are tools that let
you monkey with this as well, but I probably won't mess with that. I'm just
curious.



Hitting everything with the same sledgehammer clearly doesn't work.

And any level adjustment with require a re-encoding, which should make
crappy MP3s even more crappy. Possibly to be re-crappified again by
Bluetooth (not sure on that one) !

Is *all* the music of a similar dynamic 'nature' to start off with ? If
so, all well and good. If not, then each track (or bunch of similar
tracks) needs to be addressed separately.

geoff

[Gareth Magennis]

"Geoff" wrote in message
...

On 09/02/2016 17:34, Mike Duffy wrote:
Greetings, all.

I have a bunch of MP3 files I put on my phone (to play while in the car
via
bluetooth.)

The problem is, they came from different sources, and some are louder than
others, and it's a PITA to keep changing the volume.

I found a neat set of GNU command line audio utilities (SOX). I wrote a
messy CMD command file to extract the RMS average decibel level of each
song, then used the formula:

10 ^ ((-20 - %RMSLEVEL%) / 20) (I wrote a 'quick&dirty' c++ for this.)

to get the volume (amplitude) multiplier for each song to bring it to -20,
which is pretty close to the aggregate average of the entire collection.
Then I ran an overnight script to adjust the volume of each MP3 file as
calculated above.

Then, just for fun, I re-ran the original script that calculates the RMS
db
levels. I expected each song to be exactly -20. They are all pretty close
(around -20.5 db), so the volumes are close enough that I've solved my
problem. But I am curious. why are they not exactly at -20 db average?

There was some clipping for the songs that needed their volume increased.
Probably these were just scratches on the original vinyl. But even the
ones
where volumes were lowered did not go exactly to -20.

Am I experiencing a compression-related syndrome? There are tools that let
you monkey with this as well, but I probably won't mess with that. I'm
just
curious.



Hitting everything with the same sledgehammer clearly doesn't work.

And any level adjustment with require a re-encoding, which should make
crappy MP3s even more crappy. Possibly to be re-crappified again by
Bluetooth (not sure on that one) !

Is *all* the music of a similar dynamic 'nature' to start off with ? If
so, all well and good. If not, then each track (or bunch of similar
tracks) needs to be addressed separately.

geoff







There is software that will allow simple mp3 editing, without re-encoding to
yet another mp3 and losing even more data.
I use this freeware/donation, I assume there must be others out there.

http://mpesch3.de1.cc/mp3dc.html


Gareth.

[Trevor]

On 9/02/2016 7:24 PM, Gareth Magennis wrote:
There is software that will allow simple mp3 editing, without
re-encoding to yet another mp3 and losing even more data.
I use this freeware/donation, I assume there must be others out there.

http://mpesch3.de1.cc/mp3dc.html

Yep, mp3gain and mp3 direct cut for starters.

Trevor.

[Trevor]

On 9/02/2016 7:38 PM, Trevor wrote:
On 9/02/2016 7:24 PM, Gareth Magennis wrote:
There is software that will allow simple mp3 editing, without
re-encoding to yet another mp3 and losing even more data.
I use this freeware/donation, I assume there must be others out there.

http://mpesch3.de1.cc/mp3dc.html

Yep, mp3gain and mp3 direct cut for starters.


Whoops, I should have checked that URL, you already (sort of) said MP3
direct cut!

Trevor.

[Gareth Magennis]

"Trevor" wrote in message ...

On 9/02/2016 7:38 PM, Trevor wrote:
On 9/02/2016 7:24 PM, Gareth Magennis wrote:
There is software that will allow simple mp3 editing, without
re-encoding to yet another mp3 and losing even more data.
I use this freeware/donation, I assume there must be others out there.

http://mpesch3.de1.cc/mp3dc.html

Yep, mp3gain and mp3 direct cut for starters.


Whoops, I should have checked that URL, you already (sort of) said MP3
direct cut!

Trevor.






Just checked mp3again, and that appears to be able to do pretty much what
the OP wants.


Gareth.

[geoff]

On 09/02/2016 21:38, Trevor wrote:
On 9/02/2016 7:24 PM, Gareth Magennis wrote:
There is software that will allow simple mp3 editing, without
re-encoding to yet another mp3 and losing even more data.
I use this freeware/donation, I assume there must be others out there.

http://mpesch3.de1.cc/mp3dc.html

Yep, mp3gain and mp3 direct cut for starters.

Trevor.



mp3dc trimming and cutting sections of the song, sure. And changing
amplitude mp3gain in a linear fashion maybe.

But changing dynamics changes the actual encoded waveform significantly,
which surely requires re-encoding, just as if it was a totally different
song (which it effectively is) ?

Or does mp3gain actually somehow compress dynamics without (maybe
secretly, internally) re-encoding ?

geoff

[[email protected]]

geoff wrote: "- show quoted text -
mp3dc trimming and cutting sections of the song, sure. And changing
amplitude mp3gain in a linear fashion maybe.

But changing dynamics changes the actual encoded waveform significantly,
which surely requires re-encoding, just as if it was a totally different
song (which it effectively is) ?

Or does mp3gain actually somehow compress dynamics without (maybe
secretly, internally) re-encoding ?

geoff "


NO. And stop speculating and spreading
false information - even if it did not originate
from you. This is why, 12 years ago, so
many good folks were led to believe Saddam
Hussein bankrolled and carried out 9/11 for
cryin out loud!

Original poster:
I use mp3gain, and it does NOT futz
with the dynamic range of my mp3s.
It is simply a VOLUME LEVELER.
NOTHING MORE.

[Scott Dorsey]

Geoff wrote:
But changing dynamics changes the actual encoded waveform significantly,
which surely requires re-encoding, just as if it was a totally different
song (which it effectively is) ?

Or does mp3gain actually somehow compress dynamics without (maybe
secretly, internally) re-encoding ?

mp3gain doesn't compress dynamics, it alters the overall level at which the
track is played back. Just like Dolby AC3 files with their DIALNORM
parameter, the mp3 file has a parameter that sets the playback level. You
can set it higher or lower and change the volume without changing anything
in the encoding.

Mind you, if you play with it too much, you can find that encoding artifacts
get exaggerated. But that's just the nature of the beast.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Trevor]

On 9/02/2016 10:23 PM, Geoff wrote:
On 09/02/2016 21:38, Trevor wrote:
On 9/02/2016 7:24 PM, Gareth Magennis wrote:
There is software that will allow simple mp3 editing, without
re-encoding to yet another mp3 and losing even more data.
I use this freeware/donation, I assume there must be others out there.

http://mpesch3.de1.cc/mp3dc.html

Yep, mp3gain and mp3 direct cut for starters.


mp3dc trimming and cutting sections of the song, sure. And changing
amplitude mp3gain in a linear fashion maybe.

But changing dynamics changes the actual encoded waveform significantly,
which surely requires re-encoding, just as if it was a totally different
song (which it effectively is) ?

Or does mp3gain actually somehow compress dynamics without (maybe
secretly, internally) re-encoding ?

Nope, but I think the OP just wants similar volume levels, not dynamic
changes. If listening to recently mastered music, it will already have
all the compression you could want. You will probably find the only way
to get similar levels is to reduce the peak level of the most highly
compressed songs, since most music peaks near enough to 0dBFS these
days. MP3gain can reduce peak level without re-encoding.

Trevor.

[Mike Duffy]

On Tue, 9 Feb 2016 20:58:56 +1300, Geoff wrote:

And any level adjustment with require a re-encoding, which should make
crappy MP3s even more crappy. Possibly to be re-crappified again by
Bluetooth (not sure on that one) !

Thanks, Geoff. This confirms my limited understanding of 'level
compression'. And yes, bluetooth is crappy, and most of the MP3s were
digitized from the cheap (free?) motherboard audio chip on my PC and about
half from cassette tapes and a 'boom box' that had not seen action for 25
years. But it's for the car. Cassette hiss is virtually indistinguishable
from road noise, at least to my ears.

[JackA]

On Monday, February 8, 2016 at 11:34:06 PM UTC-5, Mike Duffy wrote:
Greetings, all.

I have a bunch of MP3 files I put on my phone (to play while in the car via
bluetooth.)

The problem is, they came from different sources, and some are louder than
others, and it's a PITA to keep changing the volume.

Ha!! Welcome to the gurus of CD mastering!!! I had to find out the underlying problem, no one here knew! Yeah, Pro they claim!...

Forget compression and all that amateur stuff. Get a copy of Goldwave. FIRST, remove VARYING DC offset, very simple. Then, under Edit menu, select the song and give it a Light Dynamic boost! Very simple. From there it's your choice.


I found a neat set of GNU command line audio utilities (SOX). I wrote a
messy CMD command file to extract the RMS average decibel level of each
song, then used the formula:

10 ^ ((-20 - %RMSLEVEL%) / 20) (I wrote a 'quick&dirty' c++ for this.)

to get the volume (amplitude) multiplier for each song to bring it to -20,
which is pretty close to the aggregate average of the entire collection.
Then I ran an overnight script to adjust the volume of each MP3 file as
calculated above.

Then, just for fun, I re-ran the original script that calculates the RMS db
levels. I expected each song to be exactly -20. They are all pretty close
(around -20.5 db), so the volumes are close enough that I've solved my
problem. But I am curious. why are they not exactly at -20 db average?

There was some clipping for the songs that needed their volume increased.
Probably these were just scratches on the original vinyl. But even the ones
where volumes were lowered did not go exactly to -20.

Am I experiencing a compression-related syndrome? There are tools that let
you monkey with this as well, but I probably won't mess with that. I'm just
curious.

Some songs need to be compressed, if not, they sound as good as dead fish compared to vinyl records!!

Jack

[Mike Duffy]

On Mon, 8 Feb 2016 23:34:04 -0500, Mike Duffy wrote:

Greetings, all.

I ... used the formula:

10 ^ ((-20 - %RMSLEVEL%) / 20) (I wrote a 'quick&dirty' c++ for this.)

But I am curious. why are they not exactly at -20 db average?

Upon reading the docs in one of the suggested utilities (mp3gain), I
arrived at the answer to my question, which nobody actually addressed.

Apparently, the MP3 format only has a resolution of about 1.5 db which is
used as the 'base' multiplier for all of the final amplitude values after
they have been computed from the other data. If the volume is increased
slowly, nothing happens until it is increased enough to bump the base
volume multiplier to the next quantum value.

Thus we should expect the resultant MP3 encoded signal to have an RMS db
level within .75 db of the requested value. The actual value depends on the
power integral of the unmultiplied signal envelope. (Excuse me if I have
used the wrong terminology. I'm not an audio engineer.)

The a/n utility also has a 'album' averaging function, which is great
because my Win7 CMD.EXE script was already a few pages long and I was
thinking about how to keep relative track volumes within an album
consistent with their original ratios.

Thanks again to all who replied. Are you all always this polite and
helpful? Over the years I've roamed a lot of usenet groups and this one
seems to be missing the usual collection of egotists, boneheads, and
deliberate trolls. Now that I've got the knowledge I came here to seek,
maybe I'll lurk a while just to see if I managed to catch everyone on a
good day or if the bozos were all hung over from superbowl parties.

[Trevor]

On 10/02/2016 4:14 PM, Mike Duffy wrote:
Thanks again to all who replied. Are you all always this polite and
helpful? Over the years I've roamed a lot of usenet groups and this one
seems to be missing the usual collection of egotists, boneheads, and
deliberate trolls. Now that I've got the knowledge I came here to seek,
maybe I'll lurk a while just to see if I managed to catch everyone on a
good day or if the bozos were all hung over from superbowl parties.

Unfortunately there are a couple of bozo's here, (and a couple just have
their bad days) but definitely not as bad as many other groups IME.

Trevor.