Hi. Why don't mp3 conversions of my recordings sound so good?

When I encode a commercial CD to mp3 (128kbps) format, I'm always
satisfied with the results. The difference I can tell between the mp3
and the 16bit/44.1kHz original is enough for me to ignore. But when I
encode MY recordings to mp3, the difference between my 16/44.1
original and my 128kbps mp3 is readily apparent to me. For example, in
the mp3, my cymbals and hihats sound mushy ("tap-tap-tap" becomes
"shtap-shtap-shtap"), I hear some flutter in long piano notes, and
other issues pop up.

But when I compare my recordings at their full 16/44.1 resolution to
commercial CDs at 16/44.1, they sound comparable. My recordings,
while not made on the best equipment, still manage to have most of the
space and have noise almost as low as some of my favorite commercial
recordings.

So what's going on here? For the record, it's always been a good test
of my recordings to convert them to mp3 just to see how much worse the
mp3 sounds than the original. I've improved my recordings this way.
But even after improving my recording methods, I think I'm still
inadvertently doing something in my recordings that is messing with
the (128kbps) mp3 encoding algorithm.

I want to fix whatever it is I'm doing wrong, but whatever I'm doing,
it's INAUDIBLE when I compare my 16/44.1 recordings to commercial CDs
at 16/44.1. Could my recordings have too much treble? Too much
hiss/information above 15kHz? Could the dither and noise shaping that
I use be throwing the mp3 algorithm off?

Thanks,
Roger

"Roger Carlson" wrote in message
m...
Hi. Why don't mp3 conversions of my recordings sound so good?

When I encode a commercial CD to mp3 (128kbps) format, I'm always
satisfied with the results. The difference I can tell between the mp3
and the 16bit/44.1kHz original is enough for me to ignore. But when I
encode MY recordings to mp3, the difference between my 16/44.1
original and my 128kbps mp3 is readily apparent to me. For example, in
the mp3, my cymbals and hihats sound mushy ("tap-tap-tap" becomes
"shtap-shtap-shtap"), I hear some flutter in long piano notes, and
other issues pop up.

But when I compare my recordings at their full 16/44.1 resolution to
commercial CDs at 16/44.1, they sound comparable. My recordings,
while not made on the best equipment, still manage to have most of the
space and have noise almost as low as some of my favorite commercial
recordings.

So what's going on here? For the record, it's always been a good test
of my recordings to convert them to mp3 just to see how much worse the
mp3 sounds than the original. I've improved my recordings this way.
But even after improving my recording methods, I think I'm still
inadvertently doing something in my recordings that is messing with
the (128kbps) mp3 encoding algorithm.

I want to fix whatever it is I'm doing wrong, but whatever I'm doing,
it's INAUDIBLE when I compare my 16/44.1 recordings to commercial CDs
at 16/44.1. Could my recordings have too much treble? Too much
hiss/information above 15kHz? Could the dither and noise shaping that
I use be throwing the mp3 algorithm off?

Just some guesses (since I obviously don't have access to any of your
mixdown files to know the specifics of them) to get the conversation going:

1.) Levels? Are your levels too low, perhaps? You might be introducing some
noise into the equation that would then have to be encoded by your mp3
encoder. Or are they too hot? Are you getting perhaps a "little" bit of
distortion that might go unnoticed until you encode it to mp3, and then it
turns into square-wave hell as a result of the reduced resolution?

2.) DC offset? Are you making sure your final mixed or mastered tracks are
free of this? If not, it could be enough to "flange" a bit once
compressed/encoded.

3.) Frequency Range? For example, do you have ****loads of content above
15k? Metric buttloads below 60 or 80? You have to remember that you must be
losing SOMETHING when converting to mp3 - otherwise how could a
10-meg-per-minute .wav file be transformed into a 1-meg-per-minute mp3?
Mp3's do lop off some high end & some low end; and IIRC, the amount varies
with the type of encoding. Some mp3 encoding software seems to work better
than others.

4.) Panning? Are you hard-panning lots of things? MP3's also tend to shrink
the soundstage laterally, so if you've got lots of things that are
hard-panned left & right, that could have some effect on it, too.

5.) Voting? Have you voted in a Democratic Primary or Caucus? If so, that
could certainly affect your mp3 conversion results (just kidding - felt I
had to interject some political content or else it might not be considered
a viable post on this newsgroup).
--


Neil Henderson
Progressive Rock
http://www.saqqararecords.com

Roger Carlson wrote:
Hi. Why don't mp3 conversions of my recordings sound so good?

When I encode a commercial CD to mp3 (128kbps) format, I'm always
satisfied with the results.
The difference I can tell between the mp3
and the 16bit/44.1kHz original is enough for me to ignore.


The problem is obviou with your acceptance of low standards.

But when I
encode MY recordings to mp3, the difference between my 16/44.1
original and my 128kbps mp3 is readily apparent to me.
But when I compare my recordings at their full 16/44.1 resolution to
commercial CDs at 16/44.1, they sound comparable. My recordings,
while not made on the best equipment, still manage to have most of the
space and have noise almost as low as some of my favorite commercial
recordings.

I would suspect that the 'problem' is that your recordings , which sound
good to you in native format, are not as compressed as the commercial ones.
They sound good to you for good reasons, the commercial ones may sound good
for other resons.

But the mp3 encoding by not seeing as many 'strident' components, is
discarding more programme as being not worthy of inclusion.

So what's going on here? For the record, it's always been a good test
of my recordings to convert them to mp3 just to see how much worse the
mp3 sounds than the original. I've improved my recordings this way.
But even after improving my recording methods, I think I'm still
inadvertently doing something in my recordings that is messing with
the (128kbps) mp3 encoding algorithm.

I want to fix whatever it is I'm doing wrong, but whatever I'm doing,
it's INAUDIBLE when I compare my 16/44.1 recordings to commercial CDs
at 16/44.1. Could my recordings have too much treble? Too much
hiss/information above 15kHz? Could the dither and noise shaping that
I use be throwing the mp3 algorithm off?

Too much dynamic range and 'delicacy'. You are doing nothing wrong. You are
demonstarting the failures in the design criteria of MP3 encoding.

geoff

"Roger Carlson" wrote in message
m
Hi. Why don't mp3 conversions of my recordings sound so good?

It's a matter of picking the right encoder, the right bitrate, and maybe
aceepting a few audible artifacts.

For example, in the mp3, my cymbals and hihats sound mushy ("tap-tap-tap"
becomes
"shtap-shtap-shtap"), I hear some flutter in long piano notes, and
other issues pop up.

That's pretty intolerable, and substandard. Find a better encoder and use
higher bitrates.

Check the Hydrogen Audio forums for recent wisdom about which are the best
encoders, today.

Its probably the fact that youve been listening to the wav so long (in
mixing) and then convert it to the MP3 with a lousy converter and you hear
all the artifacts..I get that too

Dave
"Arny Krueger" wrote in message
...
"Roger Carlson" wrote in message
m
Hi. Why don't mp3 conversions of my recordings sound so good?

It's a matter of picking the right encoder, the right bitrate, and maybe
aceepting a few audible artifacts.

For example, in the mp3, my cymbals and hihats sound mushy ("tap-tap-tap"
becomes
"shtap-shtap-shtap"), I hear some flutter in long piano notes, and
other issues pop up.

That's pretty intolerable, and substandard. Find a better encoder and use
higher bitrates.

Check the Hydrogen Audio forums for recent wisdom about which are the best
encoders, today.

"Zeppelin4Life" wrote in message
.. .
Its probably the fact that youve been listening to the wav so long (in
mixing) and then convert it to the MP3 with a lousy converter and you hear
all the artifacts..I get that too

If you use the analogy that "pro" mix is better mixed than one done at home,
and the fact that the MP3 algorithm decides what can be thrown away (what's
important and not) it can give that home mix a completely different sound
than the original file. It might be throwing away frequencies you don't want
in certain sections of the song, thereby enhancing others, etc. I've had
parts of songs that were practically inaudible in the original mix brought
way forward just by MP3 conversion (this was on a pro mix BTW). It would
skew an novice mix much more than a pro mastered mix I would think. There
are numerous options, rolloffs, etc. you can set that might minimize the
artifacts. How dynamic is your mix you are trying to encode? I highly
recommend LAME for encoding.

(Roger Carlson) wrote in
m:

When I encode a commercial CD to mp3 (128kbps) format, I'm always
satisfied with the results. The difference I can tell between the mp3
and the 16bit/44.1kHz original is enough for me to ignore. But when I
encode MY recordings to mp3, the difference between my 16/44.1
original and my 128kbps mp3 is readily apparent to me. For example, in
the mp3, my cymbals and hihats sound mushy ("tap-tap-tap" becomes
"shtap-shtap-shtap"), I hear some flutter in long piano notes, and
other issues pop up.


What MP3 encoder are you using?

Do you notice the same differences when creating MP3s of more decent
quality (the Lame encoder with the r3mix preset for example)?

Strange stuff in higher frequencies are not uncommon in MP3s, and 128kbps
is a bit too low for decent quality.

Also, some music is more hurt by MP3 encoding than other music. Are you
comparing to commercial recordings of the same kind of music as what you're
producing?

Some thoughts that may be completely off the mark (both because I don't
know anything at all about your recordings or the commercial recordings
you're comparing to, and because I'm partly guessing anyway):

* What levels are you files at compared to the commercial stuff you're
comparing to?

Some MP3 encoders have problems with low levels.

* Do you have a lot of stuff at high frequencies?

Some MP3 encoders have a rather low high frequency cutoff (I think
AudioGrabber (at least old versions) cuts everything above 15K for
example).

* Dynamic range? Details? (Compared to the commercial recordings in
question.)

This is purely a guess, I have no idea whatsoever if there's anything to
this thought: A lot of commercial music is heavily compressed, wich might
make the MP3 encoding less obvious (please correct me someone if this is
isn't so).

If your own recordings are more detailed than the commercial recordings
you're comparing to, they can be more hurt by the encoding.


/Jonas

I would suspect that the 'problem' is that your recordings , which sound
good to you in native format, are not as compressed as the commercial
ones.
They sound good to you for good reasons, the commercial ones may sound
good
for other resons.


Exactly. Not only compression but EQ and loudness maximizers are the kinds
of tools you want to use. MP3s work best from a mastered source.

blind stab...
do you have considerable (as in ANY) HF crap in there?
CRT's of any sort are ubiquitous and dump all sorts of ~15kHz crap into
mics, lines and internal cards so that it's EASY to find it on almost ANY
recording anymore. This can play HAVOC with mp3 coding. A little-discussed
thing is how processes like APOGEE's UV22 system makes recordings that can
have problems going to mp3 because of the HF dither. Likewise if you've got
stuff you can't hear floating in there at any level, it might be the
culprit.


--
John I-22
(that's 'I' for Initial...)
Recognising what's NOT worth your time, THAT'S the key.
--

From: (Roger Carlson)
Organization: http://groups.google.com
Newsgroups: rec.audio.pro
Date: 27 Jun 2004 22:09:23 -0700
Subject: Why don't mp3 conversions of my recordings sound so good?

Hi. Why don't mp3 conversions of my recordings sound so good?

When I encode a commercial CD to mp3 (128kbps) format, I'm always
satisfied with the results. The difference I can tell between the mp3
and the 16bit/44.1kHz original is enough for me to ignore. But when I
encode MY recordings to mp3, the difference between my 16/44.1
original and my 128kbps mp3 is readily apparent to me. For example, in
the mp3, my cymbals and hihats sound mushy ("tap-tap-tap" becomes
"shtap-shtap-shtap"), I hear some flutter in long piano notes, and
other issues pop up.

But when I compare my recordings at their full 16/44.1 resolution to
commercial CDs at 16/44.1, they sound comparable. My recordings,
while not made on the best equipment, still manage to have most of the
space and have noise almost as low as some of my favorite commercial
recordings.

So what's going on here? For the record, it's always been a good test
of my recordings to convert them to mp3 just to see how much worse the
mp3 sounds than the original. I've improved my recordings this way.
But even after improving my recording methods, I think I'm still
inadvertently doing something in my recordings that is messing with
the (128kbps) mp3 encoding algorithm.

I want to fix whatever it is I'm doing wrong, but whatever I'm doing,
it's INAUDIBLE when I compare my 16/44.1 recordings to commercial CDs
at 16/44.1. Could my recordings have too much treble? Too much
hiss/information above 15kHz? Could the dither and noise shaping that
I use be throwing the mp3 algorithm off?

Thanks,
Roger

LAME always sounded really harsh to me...I like Fraunhofer..

Dave
"Jonas Eckerman" wrote in message
12.194...
(Roger Carlson) wrote in
m:

When I encode a commercial CD to mp3 (128kbps) format, I'm always
satisfied with the results. The difference I can tell between the mp3
and the 16bit/44.1kHz original is enough for me to ignore. But when I
encode MY recordings to mp3, the difference between my 16/44.1
original and my 128kbps mp3 is readily apparent to me. For example, in
the mp3, my cymbals and hihats sound mushy ("tap-tap-tap" becomes
"shtap-shtap-shtap"), I hear some flutter in long piano notes, and
other issues pop up.


What MP3 encoder are you using?

Do you notice the same differences when creating MP3s of more decent
quality (the Lame encoder with the r3mix preset for example)?

Strange stuff in higher frequencies are not uncommon in MP3s, and 128kbps
is a bit too low for decent quality.

Also, some music is more hurt by MP3 encoding than other music. Are you
comparing to commercial recordings of the same kind of music as what
you're
producing?

Some thoughts that may be completely off the mark (both because I don't
know anything at all about your recordings or the commercial recordings
you're comparing to, and because I'm partly guessing anyway):

* What levels are you files at compared to the commercial stuff you're
comparing to?

Some MP3 encoders have problems with low levels.

* Do you have a lot of stuff at high frequencies?

Some MP3 encoders have a rather low high frequency cutoff (I think
AudioGrabber (at least old versions) cuts everything above 15K for
example).

* Dynamic range? Details? (Compared to the commercial recordings in
question.)

This is purely a guess, I have no idea whatsoever if there's anything to
this thought: A lot of commercial music is heavily compressed, wich might
make the MP3 encoding less obvious (please correct me someone if this is
isn't so).

If your own recordings are more detailed than the commercial recordings
you're comparing to, they can be more hurt by the encoding.


/Jonas

"Roger Carlson" wrote in message
m...
Hi. Why don't mp3 conversions of my recordings sound so good?

MP-3s of an unprocessed mike feed can sound absolutely stunning. Each
additional time you process audio leads to some point down the line when it
will start sounding really bad.

I think the most important reason for using really high quality audio gear
and converters is the fact that the resulting audio seems to be able to
withstand a lot more signal processing before it breaks. Generations of
digital signal processing are really no different than generations of analog
tape. You may not hear much difference in an original recording but a couple
generations down can really show up differences in both analog recorder and
converter quality.

--
Bob Olhsson Audio Mastery, Nashville TN
Mastering, Audio for Picture, Mix Evaluation and Quality Control
Over 40 years making people sound better than they ever imagined!
615.385.8051 http://www.hyperback.com

On 27 Jun 2004 22:09:23 -0700, (Roger Carlson) wrote:

Hi. Why don't mp3 conversions of my recordings sound so good?

...

I want to fix whatever it is I'm doing wrong, but whatever I'm doing,
it's INAUDIBLE when I compare my 16/44.1 recordings to commercial CDs
at 16/44.1. Could my recordings have too much treble? Too much
hiss/information above 15kHz? Could the dither and noise shaping that
I use be throwing the mp3 algorithm off?

I've got a noisy cassette recording that comes out really distorted
at 128k, and still substantially degraded (worse than a 'good' 128k)
when encoded at higher bitrates. I think the wideband tape noise
messes up the encoder and it tries to encode 'everything.' The less
busy and fewer signals/noises in the signal, the better mp3 encoding
should do. I would have thought dither and noise shaping would be at a
low enough level that it wouldn't affect mp3 encoding, but turn it off
and see (and tell us!) what happens.
This jibes with what other posters have said (and gives me a chance
to use the work jibe).

Thanks,
Roger

"Arny Krueger" wrote in message
...

It might be throwing away frequencies you don't want in certain
sections
of the
song, thereby enhancing others, etc. I've had parts of songs that
were practically inaudible in the original mix brought way forward
just by MP3 conversion (this was on a pro mix BTW).

This I got to hear!

It happens. I don't get out much (to my MP3 player in the car) now but the
next time I get a good a example I'll post.

On Mon, 28 Jun 2004 20:35:06 -0400, Ben Bradley
wrote:

I've got a noisy cassette recording that comes out really distorted
at 128k, and still substantially degraded (worse than a 'good' 128k)
when encoded at higher bitrates. I think the wideband tape noise
messes up the encoder and it tries to encode 'everything.' The less
busy and fewer signals/noises in the signal, the better mp3 encoding
should do.

Not related but maybe interesting: video compression for DVD from
noisy hand-held camcorders is very difficult compared to properly
lit and dollied "real" movies.

Chris Hornbeck

Roger Carlson wrote:

Hi. Why don't mp3 conversions of my recordings sound so good?

Dunno. It could be because your recordings are good.

When I encode a commercial CD to mp3 (128kbps) format,

Try 192 kbit, imo the lowest acceptable bit rate.

Roger


Kind regards

Peter Larsen

--
*******************************************
* My site is at: http://www.muyiovatki.dk *
*******************************************