MP3 bitrate for CD quality: my observations

Archived from groups: rec.audio.tech (More info?)

In another topic people were squabbling about which bitrate provides "true"
CD quality. I myself opted for 192Kb/sec but after some non-scientific
research, I've come to the conclusion that 256Kb/sec is the data rate at
which MP3 sound quality becomes indistiguishable from CD.

How did I come to this conclusion? I encoded some music works with MP3 using
a VBR (variable bit rate) MP3 encoder. Such encoders acutually vary the bit
rate to match the sonic information which has to be packed into the MP3
file. After encoding several tracks and playing them back I noticed that
very few (none actually) showed bitrates higher than 256Kb/sec at any time.
On the other hand: average bitrates of 192Kb/sec showed up on almost all
tracks, leading to the conclusion that 128Kb/sec encoding (the dominant
encoding so far for 'shared musc') is insufficient to encode the sound
without degradation in sound quality.

Actually VBR MP3 encoding would be even better for the music shared on the
Internet but very few people seem to do this at the moment, I don't know
why, since most MP3 encoders support it.

Archived from groups: rec.audio.tech (More info?)

I like VBR, but I've been having a blast lately with FLAC (Free Lossless
Audio Codec). Large hard drives make it practical (finally) to have lossless
music. A FLAC plug-in for Winamp and you're ready to go. FLAC 111.exe will
encode from WAV's and if necessary decode back to WAV. It will also function
as an external encoder for Audiograbber.

Mark Z.


"JohnR66" <nospam@att.net> wrote in message
news:%6ALd.135$xR1.25@bgtnsc04-news.ops.worldnet.att.net...
> "Henk Boonsma" <hboonsma@teranet.news> wrote in message
> news:1107184990.f502af5b38bd7e545566fed29fc19cc8@teranews...
>> In another topic people were squabbling about which bitrate provides
>> "true"
>> CD quality. I myself opted for 192Kb/sec but after some non-scientific
>> research, I've come to the conclusion that 256Kb/sec is the data rate at
>> which MP3 sound quality becomes indistiguishable from CD.
>>
>> How did I come to this conclusion? I encoded some music works with MP3
>> using
>> a VBR (variable bit rate) MP3 encoder. Such encoders acutually vary the
>> bit
>> rate to match the sonic information which has to be packed into the MP3
>> file. After encoding several tracks and playing them back I noticed that
>> very few (none actually) showed bitrates higher than 256Kb/sec at any
>> time.
>> On the other hand: average bitrates of 192Kb/sec showed up on almost all
>> tracks, leading to the conclusion that 128Kb/sec encoding (the dominant
>> encoding so far for 'shared musc') is insufficient to encode the sound
>> without degradation in sound quality.
>>
>> Actually VBR MP3 encoding would be even better for the music shared on
>> the
>> Internet but very few people seem to do this at the moment, I don't know
>> why, since most MP3 encoders support it.
>>
>>
> I used to download and encode at 128. I wish I haddn't. It has the nasty
> swishy sound to it. 160 seems to satisfy my ears. Almost never can I pick
> out that MP3 charateristic from the music at 160. I listen to a lot of
> 60's and 70's music where the quality of the recording makes more of a
> difference. If I listened to classical, or all top notch recordings, I'd
> opt for 192 or higher.
> John
>

Archived from groups: rec.audio.tech (More info?)

Well, since Zip as we know it compresses WAV files almost not at all, a 50%
compression for lossless music I feel is very acceptable, especially as I
say when large hard drives, 200 gig and up, are now available at very
reasonable prices. Also, a FLAC file can play directly without having to
"un-zip".

I once figured I could put virtually my entire music collection as WAV files
on just one of the TWO 200g drives I have in my computer.

I'm happy to archive music as VBR mp3, and not bother with lossless schemes
for material I consider lesser quality in the first place. But for higher
quality recordings that are in my current "hot" list, and for those I
consider "important" to archive with no loss, I think FLAC is ideal.

I'm not saying I could hear the difference, I'm not at all sure I could. But
there's no getting around the peace of mind factor of knowing the music file
is lossless, and if necessary can be converted back to WAV and re-burned as
a CD with no compromise in quality.


Mark Z.

"Henk Boonsma" <hboonsma@teranet.news> wrote in message
news:1107249934.59a8365d827ea5819f3e9f6a590abe70@teranews...
>
> "Mark D. Zacharias" <mzacharias@yis.us> wrote in message
> news:3684toF4vqh1eU1@individual.net...
>> I like VBR, but I've been having a blast lately with FLAC (Free Lossless
>> Audio Codec). Large hard drives make it practical (finally) to have
> lossless
>> music. A FLAC plug-in for Winamp and you're ready to go. FLAC 111.exe
>> will
>> encode from WAV's and if necessary decode back to WAV. It will also
> function
>> as an external encoder for Audiograbber.
>>
>
> The compression ratio of FLAC is about 50% which means an average CD song
> (50MB) will still compress to a sizeable 25MB. Also, FLAC doesn't
> (contrary
> to the claims) compress .WAV files any better than .zip.
>
>
>
>

Archived from groups: rec.audio.tech (More info?)

"HP" <x@x.invalid> wrote in message news:cto7k201cv1@news1.newsguy.com
> Peter <usenet_AT_tecno.demon.co.uk@junkblock.com> wrote:
>> I tried this test encoding a couple of tracks using LAME at 128kb/s.
>> I then randomly arranged the tracks on CD and gave them to a couple
>> of friends, one of which was a BBC sound engineer.
>>
>> For good measure I included on the CD encoding's at 160,192,256 as
>> well.
>>
>> Playback was via a NAD3020 into Chartwell LS3/5a's
>>
>> Result : They failed spectacularly to identify the original WAV on
>> the CD from the MP3 encoded version. Nor even could they identify
>> between the various MP3 formats.
>
> All these encoded versions, did they all came from the same
> soundtrack? And, did your friends hear the original WAV beforehand?
> Otherwise, depending on the music contents (specially in the case of
> pop/synthesized music) they might not be able to tell whether the
> artifacts were part of the original song or not.
>
> And, they might not have been able to recognize the original WAV,
> even less indetify the different MP3 formats, but I bet all they
> complained that all the recordings did sound *different*.

Most if not of your concerns can be addressed by use of this testing
methodology:

www.pcabx.com

Archived from groups: rec.audio.tech (More info?)

>...But what upsets me the most from listening to MP3
>audio is that I get premature ear fatigue, even after only ten minutes of
>listening, and for some reason very tormenting

This 'fatigue' is a variant observation to that I have often heard of
the objection not just to MP3, but to ANY compressed audio technology
(as if 16bit sampling was not, in itself, compression!)

The problem I have is that when I have run tests (see earlier in this
thread) even the expert sound recordists can't identify better than
chance when exposed to blind comparisons of uncompressed and
psychoacoustic methods for compressed audio.

The previous author claims he can hear (or in some way discern,
through fatigue) the difference. I would therefore love to run the
same test to qualify the response.

Archived from groups: rec.audio.tech (More info?)

Peter <usenet_AT_tecno.demon.co.uk@junkblock.com> wrote:
: This 'fatigue' is a variant observation to that I have often heard of
: the objection not just to MP3, but to ANY compressed audio technology
: (as if 16bit sampling was not, in itself, compression!)

16-bit sampling is not the same kind of compression, I cannot totally agree
with you there, sorry. You are not compressing, but digitizing. Yes, you end
up with less information than the original analog waveform have, you
introduce quantization noise, both in amplitude and phase, and yes, you are
masking very small noises and/or music passage levels, but that is not the
same selective masking that MP3 does. Neither you are replacing two closely
spaced frequencies with a single tone. Granted, 16-bit sampling will never
sound as good as a 30-ips tape, but it will sound better than MP3
compression.

About the ear fatigue, you are right. Other compression methods, like ATRAC,
actually tend to cause the same fatigue, at least on me. And it is *not* an
exclusive effect of a given compression method, as I tend to get ear fatigue
also from using low quality amplifiers, poor speaker systems, bad headphones,
poor recordings. etc... you get the point.

One thing I have noticed over the years is that different people tend to
concentrate their attention on different aspects of the music, and that not
everybody hears the same. If neither you nor your test subjects get ear
fatigue with MP3, then lucky you, but do not assume that other people won't.
Same goes for the compression artifacts. After all, not every person can
distinguish the sound of a Stradivarius from an ordinary violin.

Here goes a little story: a friend gave me a couple of years ago a nice
portable Sharp MD-MT200 mini-disc player together with several disks. It
uses the ATRAC technology. I liked it a lot: small, portable and with discs
that would fit in a pocket. And when I first listened to it, it sounded near
hi-fi. It had the usual "enhanced" stereo mode, extra bass and all that
stuff to make it to sound better on headphones (not always the case). So I
decided then to record a few of my CDs on mini-discs. While I was doing that
I was monitoring the recording with the headphones and I was thinking to
myself "this thing can actually sound beautiful". I used the lowest possible
compression ratio (5:1) for the highest quality possible. But the surprise
came later when I listened to what was actually recorded: yes, it sounded
still rather satisfactory but it was not the crystal-clear sound from the
original CD, somehow the instruments did not sound the same. Somewhat opaque
sound, undefined and unclear here and there, fuzzy and with much less stereo
3D imaging (where the instruments are now?). I thought "I could live with
this anyway, I can still use it when traveling, since it's so small", so I
started using it more regularly for a couple of days or so. Or, I should
say, tried. That thing produced one of the most annoying ear fatigues I've
ever experienced! It was the kind of fatigue that makes you want to take out
the damned headphones and throw the whole thing to the opposite corner of
the room. Now I understand why the person that gave it to me did it for
free, despite the relatively high original price he paid for it. Now it is
sitting at home somewhere in a drawer, I just can't put that earphones on
again. I was thinking to give it away to someone else, but I feel kind of
guilty about doing that.

Something similar I experienced with MP3. I thought at first "finally a good
compression method with inaudible loss of quality, now I can start
transfering my tapes, LPs and CDs to the computer without running out of
disk space." But then again, the deception: the "inaudible" was actually
audible. I tried increasing the bitrate, played with different encoders and
decoders (BTW, they tend to make much more difference in quality than the
bitrate itself). Sometimes I got better results, sometimes worse. But never
the claimed "inaudibility" of the compression. After a while I could even
tell what I had encoded with which encoder. Never got the same crystal clear
sound as the original had. And I never could listen more than half hour to
those MP3. Too much buzzing in the middle of my head, never as clear as I
wanted. Again, some recordings suffered much less than others, a few sounded
very acceptable, but the problem is that I do not listen to the same thing
over and over.

And, another thing: after a while my ears started adapting to the MP3 sound,
but not in the way I would have liked. I started actually hearing MP3-like
artifacts in my original tapes and even when listening to live concerts! It
was a feeling like if my brain actually *expected* the artifacts to be there
at a certain moment and reproduced the sensation. It happened mostly when I
was relaxed enjoying the music, and was a rather disturbing experience.

: The problem I have is that when I have run tests (see earlier in this
: thread) even the expert sound recordists can't identify better than
: chance when exposed to blind comparisons of uncompressed and
: psychoacoustic methods for compressed audio.

Tests, tests... Try to listening one complete day only to MP3 recordings.
And try the next the originals.

Listening experiences vary a lot from individual to individual, and depend
considerably from the equipment used, quality of the original recordings,
listening room characteristics, and the duration of the hearings. I don't
have any details about how do you perform the tests (other than you are
apparently using the ABX program to collect the results), so I cannot take
your results as a proof that MP3 recordings have the same quality as a CD
one.

Very often it can take days before you notice what's wrong with a given
recording or piece of equipment. Did ever happened to you that after buying
an amplifier (or speakers, or headphones) because it sounded good the first
day you actually started hating it after a week of listening to it due to
its poor sound quality?

: The previous author claims he can hear (or in some way discern,
: through fatigue) the difference. I would therefore love to run the
: same test to qualify the response.

And I would love to try the test. I'm still hoping to find a MP3 that I
could actually enjoy listening to.

Regards,
Hector.