Skip to main content
Topic: MP3 VBR Transparency Rate (Read 12116 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

MP3 VBR Transparency Rate

Hi, new to the forums as well as to the audio world.

I've recently discovered the availability of lossless formats such as FLAC and APE (I know it's a bit late) and have since reripping all of my songs into such formats. I always thought that using lossless formats in my MP3 player is the best way to go, until I took an ABX test recently. I realised that I had real difficulty distinguishing Lossless formats with MP3 format at VBR, even at ~120kbps (encoded with LAME -V7 or -V6). This kind of hit me like a tornado, I mean, I can't even tell the difference of compressed songs at such a low bitrate. Perhaps it is due to my ordinary soundcard and headphones (both are stock materials)...... I don't know.

What I'm wondering is that, what is the transparency bitrate for VBR MP3 for most people in here? I am starting to consider removing all of my FLAC files from my Sansa Clip and start compressing them into MP3. Please post honestly, as I've read through a few threads here where people claim to be able to tell the difference between 320kbps and the lossless file, which I think is ridiculous.

MP3 VBR Transparency Rate

Reply #1
There are some samples which are ABXable to a small group of people even at 320kbps. Just because they can means that you can too. Since you have difficulty distinguishing -V6 from the original, just use -V5 or -V4 just to be on the safe side.

For most people here, transparency starts at around -V5.

MP3 VBR Transparency Rate

Reply #2
... What I'm wondering is that, what is the transparency bitrate for VBR MP3 for most people in here? ...

If you look at last fall's 128 kbps mp3 listening test (see the listening tests forum) you will find that all of the mp3 encoders tested yielded a perfect or close to perfect quality to the average of listeners on the average of samples. In the case of Lame 3.98.2 that was at a setting of -V5.7.
So in a practical sense you're fine with a VBR setting yielding a bitrate of 128 kbps on average.

Taking a closer look you'll find that what is valid for the average of samples isn't true for any sample. There were samples in the listening test that provided problems for the one or other encoder, and the quality in these cases was lower to essentially lower than what was the case in the average view.
That's why many mp3 users allow for a security margin and use a higher bitrate. Quality improves with those hard-to-encode samples and can become perfect.
Ratio of quality vs. bitrate increase is better in the moderate bitrate region, that's why according to last year's poll most Lame users prefer a Lame VBR setting around -V2 (which means an average bitrate of roughly 200 kbps or a bit lower).

On the other hand many users don't have storage space problems and can allow for -V0 (roughly 240 kbps on average). In fact this group took second place in last year's poll. With such a setting you are very much on the safe side (though there are still samples not encoded perfectly). If 240 kbps is too much you can reduce average bitrate by using mp3packer afterwards (a lossless and very fast process) and/or restrict the extreme high frequency area to 17.5 or 16.7 or 16.0 kHz or in the variable way of the -Y option according to your likings. (You'll probably find that you don't miss anything beyond 16.0 kHz. -V0 --lowpass 16.0 and the mp3packer procedure yields an average bitrate around 215 kbps).

All depends on how much you care about problematic samples. Probably it's wise not to do too much (there's a chance you personally never encounter one), but maybe it's also wise not to ignore them completely. It's up to you where to put your personal sweet spot.

Maybe a rough overview in problem areas can help you decide how much to care about problems:
- electronic music
Some of the worst problems arise from this genre, especially when it's about sharp attacks of impulses artificially produced here. If you are a lover of this genre you should be problem-aware up to better not use mp3.
- harpsichord music.
Very serious issues for mp3 can arise from harpsichord music. With very high quality settings however quality of a good mp3 encoder should be very acceptable at least.
- long-stretched tones of whatever origin
Long-stretched tones can produce distortions and/or a tremolo-like effect when using bitrates too low.
- Voices
Voices can be encoded with subtle distortions when using a bitrate too low. Usually voices are encoded fine even at low bitrate, but it's not always like this. To me french woman singers are pretty prone to this issue, but last night when I encoded african music at moderate bitrate I ran upon this issue also with male african voices. I guess we're very sensitive to voice, so errors at the same subtle error level are recognized easier than when originating from instruments.
lame3995o -Q1.7
opus --bitrate 140

MP3 VBR Transparency Rate

Reply #3
These polls might help you decide.

MP3 VBR Transparency Rate

Reply #4
I never use anything lower than V4. Even in the old days; preset medium.

My take on this is to work out the typical bitrate you can tolerate regarding storage and to lower bitrate with lowpassing rather than more aggressive methods like V7 / V6

It like Halb27 said;

Say V4 yields 160 k on average. If I wanted something closer to 135k I would rather -V4 --lowpass 15 than V5.5
For V2-V1-V0 ; Use -Y or lowpass 16 for high quality with lowered bitrate or --lowpass 15 for portable use.

On loud rock / pop for portable:

-V2 - 240k
-V2 -Y or  --lowpass 16 - 180k
-V2 --lowpass 15 - 160k

The bitrate is drasticaly lowered but the quality techniques are untouched, Thus you have lower change of artifacts than lowering than using V4. This is because humans are much less sensitive to HF content in music than artifacts in lower frequencies which are a risk when using lower V setting.



wavpack 4.8 -b256hx6c

MP3 VBR Transparency Rate

Reply #5
If I were you, I'd backup the flacs for safe-keep(archive), don't get rid of them completely..You can always encode from the flac files to choice of compression format, at a later time..

MP3 VBR Transparency Rate

Reply #6
Basically , I would stick to what we know through listening tests. V5 - V5.5 (135 k) yields decent quality for many typical uses. To lower the bitrate use --lowpass 15 -V5.5 (120k)
wavpack 4.8 -b256hx6c

MP3 VBR Transparency Rate

Reply #7
In a blind test V5 is totally transparent for me.  And I've done the test using "audiophile" grade hardware.

Let it be known that it's a typical audiophile response to say that if you can't tell the difference between MP3 and FLAC then you need to upgrade your sound card, your amp, your cables, headphones, speakers, etc.  Since none of these people are blind testing I suggest you treat all claims with skepticism.  Modern sound cards (even onboard sound on nice motherboards) measure quite well and sound very good.

There are people who can ABX 320mbps mp3s and FLAC.  There are also people who can win a Olympic gold medal for swimming.  Maybe the gold medalist needs a $1000 shark skin swin suit, but I assure you that the average person does just fine with a pair of $10 swimming shorts and no suit (short of a bionic one) will help them win an olympic gold medal. 

Same logic applies here. Just because some people can ABX some samples at high bitrates that doesn't mean FLAC should be used exclusively.  Use what is transparent to you.

I have my music in FLAC and MP3.  I keep it in FLAC for archival purposes so I have a backup of my CDs.  It's also easier to convert to other formats (mp3, ogg) that way.  I encode all my music at -v0 but it's more for psychological security as V5 is transparent to me.

MP3 VBR Transparency Rate

Reply #8
What I'm wondering is that, what is the transparency bitrate for VBR MP3 for most people in here?

-V5. I generally don't concern myself with achieving transparency 100% of the time, as audible artifacts are not usually annoying to me at -V5 (if audible at all, of course). As such, it's exclusively -V5 for me.

MP3 VBR Transparency Rate

Reply #9
V5 is abxable on every random sample from my collection - even if some isolation and concentration is needed. In most cases it sounds excellent but less so on more difficult samples. With todays storage (even yesterdays) V4 is better insurance even though lower quality settings can be adequate. For *me* to bring the bitrate to V5 levels I would --lowpass 15 -V4 and quality will be safer than V5 .
wavpack 4.8 -b256hx6c

MP3 VBR Transparency Rate

Reply #10
As for -V4 and the 15 kHz lowpass shadowking mentioned:
Due to storage space restrictions with my new DAP I am just re-encoding my classical music using Lame 3.98.2 -V4 --lowpass 15.2 (should be identical to --lowpass 15). Quality is great - perfect so far I've tested, and average bitrate is a low 137 kbps (measured not with classical music but my standard test set of various pop music).

When using such a lowpass we should keep in mind however that Lame 3.98.2 resamples to 32 kHz. Which has the advantage that encoding is more efficient saving bitrate for the same quality level compared with a 44 kHz encoding. Which has the disadvantage that eventual impulse problems are getting more severe. Lame 3.98 is relatively good at impulses even at 32 kHz sampling, so this disadvantage is not relevant to many kind of music.
lame3995o -Q1.7
opus --bitrate 140

MP3 VBR Transparency Rate

Reply #11
This kind of hit me like a tornado, I mean, I can't even tell the difference of compressed songs at such a low bitrate. Perhaps it is due to my ordinary soundcard and headphones (both are stock materials)...... I don't know.


Actually, many modern day lossy encoders can perform great at around the 128kbps bitrate.  iTunes AAC, Nero AAC, and Lame mp3 have been able to provide transparent results at around 128kbps for a long time now for many people.  I wouldn't worry about your equipment or your ears.  Many of these modern day lossy encoders have been vastly improved over the years.  Some people still look negatively at 128kbps bitrates (even with modern day encoders) but they are simply carrying over these stigmas from when encoding mp3 files was still a rather new procedure.

I use the Nero AAC encoder for my lossy need at the -q0.50 bitrate setting.  -q0.45 is able to provide transparent results but I went up to -q0.50 as my files didn't really increase by much.  There was also a Nine Inch Nails song in which -q0.45 struggled with yet -q0.50 produced transparent results for me.

However, back when I was using Lame (and from my recent tests), I determined that (for the most part) -V 2 with Lame 3.98.2 was transparent for me.  I also found out that -V 3, -V 4, and -V 5 were transparent when it came to casual listening.  My only issues with -V 4 and -V 5 were that they didn't produce transparent results in my car (yes, I brought my notebook in my car and hooked it up via an analog auxiliary input and conducted a series of blind ABX tests) until I started driving on the highway, then everything was transparent (I wasn't driving but you get the idea).

If I were going to use Lame, I would probably go with -V 3 simply because the file sizes produced with -V 2 are rather big (my library consists of some hard rock with a bunch of metal) and I hardly ever listen to music to scrutinize its quality (unless I am conducting blind ABX tests).  I will almost always listen to music to actually enjoy it (I am not saying that people here don't do that, we all care about the music or else we wouldn't be here).

MP3 VBR Transparency Rate

Reply #12
Gee guys, thanks for the many responses.

I have read through all of the comments, and learned a bit more about lowpass. I've actually heard of it, but never really used it to encode my MP3s. I just used the standard -Vx. I think I will reencode my songs using -V2, although I believe many songs will be probably be transparent to me at -V6. Just to be on the safe side. I'm doing this for a few reasons, but mainly because of psychological and the fact that my Sansa Clip has pretty much space (4GB). Thanks for sharing.

Oh, and odigg, what you said about the swimsuits is a good analogy.

MP3 VBR Transparency Rate

Reply #13
I am (slowly) working on a set of comparisons from my CDs to various VBR bitrates, and hope to update as I complete each set of tests.  So far, results of testing at 128k are encouraging me to believe that I can hear the difference on good recordings (see this thread) but I'm not nearly done with samples at 128k yet.  The differences are small enough that I expect to have serious trouble ABXing 192k.  I've been typically encoding at either 192k or 256k for serious listening on the iPod depending on the perceived quality of the recording, without the benefit of testing.  It would be nice if I can continue at 192k, should my testing conclude that it's transparent for me.

MP3 VBR Transparency Rate

Reply #14
I use FLAC primarily, for peace of mind. I have ABXed MP3 at 320kbps on some samples. However, when I do use MP3, I use V4 for noisy environments (and don't usually notice even when using nice equipment in quiet environments) and V2 when I'm concerned about quality. V0 feels like overkill to me. V5 is probably fine as well, but it doesn't handle quite as well as V4 when using nice equipment in quiet environments.

@GregDunn: Note that VBR is quality-based, not bitrate-based, and that bitrates can vary widely. Referring to quality settings by corresponding bitrates is somewhat nonsensical.

MP3 VBR Transparency Rate

Reply #15
@GregDunn: Note that VBR is quality-based, not bitrate-based, and that bitrates can vary widely. Referring to quality settings by corresponding bitrates is somewhat nonsensical.


Understood.  Should I be referencing the files in a different manner?  My encoder only says the approximate bitrate when encoding (not using LAME, because I encode with either Audion or iTunes), so I don't know how else I should distinguish them for comparisons; there's no "-Vn"  parameter equivalent.  I mean, I know I can replicate the settings, but they may not be meaningful to others who use a different system.  Since I'm encoding for my own use, I didn't see that as a problem, but I recognize that my results might not be as useful to others.

MP3 VBR Transparency Rate

Reply #16
I think referring to the encoding settings is typically the best way for others to relate to your testing and your results. Since the actual average bit rate can be highly variable, referring to any VBR file by its bit rate (i.e. 256 kbps for LAME V0) isn't especially useful unless you're referring to specific tracks or collections of tracks.

MP3 VBR Transparency Rate

Reply #17
V3 and V2  is highly likely to be transparent for most people with most of their music. But sadly if you are sensitive to precho and listen to electronic music or like drums (i find drum smearing on V2 to be quite common for me) then it can be abit annoying at those settings. Also i find V0 to be bloated and not worth the extra 50kbps for a small improvement on problem samples, even if it does fix most of my drum smearing issues.
"I never thought I'd see this much candy in one mission!"

MP3 VBR Transparency Rate

Reply #18
About the SanDisk Sansa Clip: My wife is very happy with the lame V4 encodings on hers, and she has far better hearing than I have. For me transparancy starts at V5 / V4, but as a standard I encode at V2 to be on the safe side. Apart from the lossyFlacs..

MP3 VBR Transparency Rate

Reply #19
On the other hand many users don't have storage space problems and can allow for -V0 (roughly 240 kbps on average). In fact this group took second place in last year's poll. With such a setting you are very much on the safe side (though there are still samples not encoded perfectly). If 240 kbps is too much you can reduce average bitrate by using mp3packer afterwards (a lossless and very fast process) and/or restrict the extreme high frequency area to 17.5 or 16.7 or 16.0 kHz or in the variable way of the -Y option according to your likings. (You'll probably find that you don't miss anything beyond 16.0 kHz. -V0 --lowpass 16.0 and the mp3packer procedure yields an average bitrate around 215 kbps).
With any decent LAME encoding, especially VBR, MP3packer will hardly save you anything at all, 1 kbit/s at most.  You will find that only less efficient MP3 codecs that make little to no use of the bit reservoir at CBR encodings, may benefit substantially from an MP3packer treatment.  Only then, of course, those encodings will no longer be CBR.
Check the wiki article for more background.

 

MP3 VBR Transparency Rate

Reply #20
... With any decent LAME encoding, especially VBR, MP3packer will hardly save you anything at all, 1 kbit/s at most. ....

a) The slow mp3packer -z process is indeed useless when using Lame.
b) the very fast mp3packer process (without options) is also useless when using low bitrate Lame encodings.
c) When using Lame 3.98 -V0 or similar, ABR very high bitrate or CBR 320 the very fast mp3packer process often saves considerably more than 1 kbit/s.
Bitrate saved depends heavily on the amount of 320 kbps frames used.
Whether it's worth while is nonetheless a question of taste. I personally use it even with -V1.5 though in this case advantage is pretty low.
But as it's so extremely fast I simply do it by using a .bat file containing the Lame, mp3packer (and mp3gain) process.
lame3995o -Q1.7
opus --bitrate 140

 
SimplePortal 1.0.0 RC1 © 2008-2020