Skip to main content
Topic: LAME @ 320 Vs. Other mp3 Encoders @ 320 (Read 18914 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Has anyone done any testing on only 320 bitrate encoding for LAME Vs. other encoders like Blade, FhG, Xing etc. ?


LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #2
320kbit is called "insane" for a reason in the lame presets (it was earlier even called "braindead"). Better either go down to -V0/preset extreme...... or go lossless. 320kbit is bloat for listening but not enough to make the audio technically the same as the original.

To answer your question - never heard of any blind test done at 320kbit..... even at around 200kbit it becomes difficult for people with golden ears.

- Lyx
I am arrogant and I can afford it because I deliver.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #3
But it should be possible with killer samples.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #4
I believe the question was LAME (presumably API) vs. other (Possibly inferior) MPEG-1 layer 3 implementations, not 320 Kb/s vs. original.

With that said, best wishes to those who try.
I'm too disinclined to try locating one of the old, low quality codecs to see how feasible it is.

Edit: err.. Not that I use API or anything..

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #5
Yes i meant LAME@ 320 Vs. other encoders @ 320 and ofcourse you would compare them individually to the original wave to find out which encoding is closest to it.

I don't mean any old encoders but the latest from all the other encoders like Blade, FhG or the one which comes within iTunes.

I was just reading on the net and many tests have shown that iTunes encoder is better than LAME.

I have no idea what ABX is and how to really test other than purely just listening to the encoded files and comparing them to the original.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #6
Quote
Yes i meant LAME@ 320 Vs. other encoders @ 320 and ofcourse you would compare them individually to the original wave to find out which encoding is closest to it.

I don't mean any old encoders but the latest from all the other encoders like Blade, FhG or the one which comes within iTunes.

I was just reading on the net and many tests have shown that iTunes encoder is better than LAME.

I have no idea what ABX is and how to really test other than purely just listening to the encoded files and comparing them to the original.

Which tests? The only empirically valid, double-blind test I know of that used iTunes found it to be the  worst modern MP3 codec @ 128kbps. Though quality isn't necesarily a linear function when it comes to audio codecs (e.g., a codec's performance at X bitrate doesn't always correlate to its performance at Y bitrate), I would surmise that iTunes MP3 encoder would be much worse than LAME at high bitrates as well.

Also, audio quality cannot be determined by comparing wavs in a spectrum analyzer, or doing casual A/B comparisons: the only empirically-valid, scientific way of doing this is through a double-blind (placebo-controlled) test, known as ABX. This way, you have no idea of what you are listening to and therefore your ratings cannot be colored by placebo effects, i.e. "warm, fuzzy feelings" for a certain bitrate or format.

(Edit: fixed URL)

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #7
Quote
I don't mean any old encoders but the latest from all the other encoders like Blade, FhG or the one which comes within iTunes.

blade is no longer developed and is commonly known to of far worse quality than e.g. lame.

Quote
I was just reading on the net and many tests have shown that iTunes encoder is better than LAME
any links to these tests?

Quote
I have no idea what ABX is
look here.
Nothing but a Heartache - Since I found my Baby ;)

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #8
Quote
I was just reading on the net and many tests have shown that iTunes encoder is better than LAME.

iTunes AAC-Encoder OR mp3-Encoder?

Their AAC-Encoder seems to be one of the best AAC-encoders (or maybe even the best, currently).

However, their mp3-encoder appears to be one of the worst current mp3-encoders, from what i've seen.

- Lyx

edit: yes, there is a possibility that the above is not a coincidence :-)
I am arrogant and I can afford it because I deliver.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #9
Quote
Better either go down to -V0/preset extreme...... or go lossless. 320kbit is bloat for listening but not enough to make the audio technically the same as the original.

This isn't really directed at a specific person, since I've heard this explanation way too many times lurking around here.  Yes, for most tracks there is no perceivable difference between -APE and -API, does this mean you shouldn't use the insane preset? 

No.  I don't understand what is wrong with being sure to a higher degree of certainty that your tracks will be transparent, even if only rare problem cases.  I believe 'you mind as well just go lossless' line of thinking is flawed. 

Let's say I average 900 kbps encoding my collection in flac and 320 kbps in API vs. 200 kbps for -APS.

% Difference File Size (vs. -APS):
250 kbps - 25% (Extreme)
320 kbps - 60% (Insane)
900 kbps - 450% (Flac)

That's quite an enormous gap between extreme and lossless...your telling me that it's not possible to have enough hard drive space to encode at 320 but too little to encode losslessly?

For the record...I use my computer as my home stereo and all my ripped cd's are AAC transcoding preset.  (~310-330 kbps).  Can I hear a difference between this and standard... I doubt it.  But was the extra space worth piece of mind?  I think so.  As a bonus...should I ever need to transcode these tracks to a much smaller file for portables, I can do so with an acceptable loss of quality.  At -APS this wouldn't go over nearly as well.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #10
Quote
Quote
Better either go down to -V0/preset extreme...... or go lossless. 320kbit is bloat for listening but not enough to make the audio technically the same as the original.

This isn't really directed at a specific person, since I've heard this explanation way too many times lurking around here.  Yes, for most tracks there is no perceivable difference between -APE and -API, does this mean you shouldn't use the insane preset? 

No.  I don't understand what is wrong with being sure to a higher degree of certainty that your tracks will be transparent, even if only rare problem cases.  I believe 'you mind as well just go lossless' line of thinking is flawed.


Short explanation:
- "preset medium" is the preset at which normal music sounds transparent for joe-average
- "preset standard" is the preset at which normal music sounds transparent for people with golden ears
- "preset extreme" is for people with golden ears who want to have "more than enough" (the extra headroom)

So, by using "extreme" you ARE already using the extra headroom. If you don't even trust this preset, then you basically don't trust lossy codecs in general. In that case, lossless-codecs are a better choice.

Basically, its like this - lossy codecs for listening, lossless codecs for archiving and people who don't trust lossy codecs.

"Perceived transparency" is already archived at "preset standard". However, because of the working of the mp3-format, no mp3 can make a file "technically the same as the original". So, 320kbit is overkill for listening and "not enough" for making the audio "technically the same as the original". So, its a disadvantageous choice for listening AND archiving.

Another point may be interesting. When you read threads in which listening tests happen, then you see samples where preset standard or (more rare) extreme have a problem. However, whats not so obvious is the conditions under which such tests happen:
- in those tests "killer samples" are used. This means that from hundreds of songs a few short samples are collected where a difference can be noticed.
- even with those "killer samples" the people who attend such tests often need a "training-phase" to notice the difference reliably. So, they listen to the original and the mp3 version of the sample multiple times "to get a feel of them" <--- Compare these methods to normal music listening.


Quote
Let's say I average 900 kbps encoding my collection in flac and 320 kbps in API vs. 200 kbps for -APS.

from my experience with FLAC, it doesn't average at around 900kbit...... but instead more at around 700-800kbit. True, lossless files are bigger - but i guess someone who has room for 320kbit mp3 has a big HD ?


Quote
For the record...I use my computer as my home stereo and all my ripped cd's are AAC transcoding preset.  (~310-330 kbps).  Can I hear a difference between this and standard... I doubt it.  But was the extra space worth piece of mind?  I think so.

As i mentioned already - it seems that you mistrust lossy-codecs in general, but think that by going maximum bitrate the disadvatages of lossy codecs magically become nonexistent. If you want placebo-magic, then lossy codecs are the wrong choice - because they are meant for "how the music sounds", not "how the music/bitrate looks like".

Quote
As a bonus...should I ever need to transcode these tracks to a much smaller file for portables, I can do so with an acceptable loss of quality.  At -APS this wouldn't go over nearly as well.

Well, exactly for transcoding and archiving purposes lossless-formats exist. As i said, lossy is meant for listening, not for transcoding. Lossless instead is meant for just that: having the audio technically as a perfect copy compared to the original.

Anyways, it seems that most of your reasoning comes from placebo-magic. The reason why you do the things the way you do them is not "because you can hear it", but instead "because it just feels better". In that case, every discussion about mp3 has no foundation, and lossless is instead the only logical step.
Lossy is about what one can actually hear - your reasons are not about what you can actually hear -> lossless

- Lyx
I am arrogant and I can afford it because I deliver.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #11
I messed around with blade and old xing a while back. They failed badly even past 192k on tracks with sharp precussion.
wavpack 4.8 -b3x6c

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #12
Beware of 320 kbps MP3's. 

If you do ABX testing you will probably find that LAME --preset standard is sufficient.

Should you a some point deside to use a MP3 player you will benefit from not having encoded your files at 320 kbps. It can hold more songs and battery life is extended.

If you are going to play 320 kbps MP3s on a 3rd generation Apple iPod the sound will be badly distorted. 

This is true for LAME 3.90, 3.93, 3.95, iTunes 4.2.

(I have not tested LAME 3.96, iTunes 4.5 nor the iPod mini.)

Ironnically I started encoding at 320 kbps for much the same reasons that you mention, only to find out that I had severe sound quality problems on the iPod. 

Now I encode to lossless (FLAC) and can automatically convert my lossless files to lossy. In this way I get the best available quality at any given time. When an improved encoder comes along I can automatically reconvert all my files without having to rip my CD collection all over again. 

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #13
Lyx:

I understand what you are trying to convey, and I agree on most points.  However, I was not trying to argue whether a difference can be heard between APE and APS ect... If I were trying to prove they were not sufficient I would have included ABX test results to show it.  I was simply pointing out that choosing the maximum preset for the codec should not be immediately greeted with 'you shouldn't use this bitrate'  just because you consider it overkill. 

It all depends how you define margin of safety.  Maybe the community thinks that say an extra 20 kbps is enough.  Someone else might look at that and say..."Well I do trust their opinion...but I regard myself as having pretty good hearing, and my listening equpment is well above average... so maybe an extra 40 kbps is a good idea, just to be safe."  Is this person right or wrong?  Can they percieve a difference or not?  The answer:  There is no way of telling by their response...maybe it's placebo, maybe it's not.  Does it mean choosing 320 kbps is bad?  No.  The only disadvantage is increased HD usage...and if they can spare the extra 22% over extreme... then why do you care?  It's not like the community is infalliable either...remember when r3mix was 'transparent'?...remeber when joint stereo was 'bad'?.

Not allowing people to deviate slightly from the suggested norm is not a good idea.  Consider this scenario...what if instead of following the good advice of people on this forum I decided to follow what the media was telling me (via ads, articles in magazines, bestbuy salesmen)?  All of a sudden our point of reference shifts...now a 112 kbps WMA is 'CD-Quality' and 96 kbps is good enough for joe average listener and 128 kbps is for audiophiles and 192 kbps is overkill.  This is certainly what microsoft would have us believe...or sony with their ATRAC...and Apple with their 128kbps AAC ect...).

Basically, I don't see why overkill is so horrible and to be avoided at all costs.  CDaudio is overkill, the number of simultaneous colors displayed on this screen is overkill, the strength of the suspension arm in your car is overkill.  Hell, does the fact that your car components are far stronger then they need to be bother you?  Anything more than the absolute minimum required strength for these components to survive is just for placebo effect after all.  The phrase 'What if' certainly becomes more meaningful in this scenario.  "What if I drive into a tree?"  "What if I get hit by a train?"

Why can't I demand a greater margin of safety in my encoding?  Even if it is entirely for 'piece of mind'.  Especially when I browse over to the 3.96 vs. 3.92 thread and find that there is possible degredation in these presets you refer to previously as though they are absolute and irrefutable.  I just do not have the time or patience to ABX everything I encode just to be sure that APE is sufficient...to me the transcoding preset is well worth the lost agrivation.

My trust in lossy codecs completely depends on the manner with which they are utilized.  That said, if I need something from someone and I am not sure of their level of competance, I'd much rather go with the 'insane' choice.  I suppose you could say my 'degree' of trust is lower than yours...this is not a case of yes/no, black/white, bush/gore.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #14
Useless Engineer:

Quote
Someone else might look at that and say..."Well I do trust their opinion...but I regard myself as having pretty good hearing, and my listening equpment is well above average... so maybe an extra 40 kbps is a good idea, just to be safe."


My point is that you will NOT be safe:
- Even at 320 kbps you might be able to find tracks where you feel a difference between CD and 320 kbps on your type of equipment.
- A problem like bad distortion on 3rd generation iPod turns up.
- ...

If you want to be safe then you should consider using a lossless scheme like FLAC.

Quote
My trust in lossy codecs completely depends on the manner with which they are utilized. That said, if I need something from someone and I am not sure of their level of competance, I'd much rather go with the 'insane' choice. I suppose you could say my 'degree' of trust is lower than yours...this is not a case of yes/no, black/white, bush/gore.


Encoders improve as time goes by and new versions are released and validated through listening tests. If you choose one of the newer encoders you should be prepared to reconvert your files as bug fixes come along.
By ripping to lossless you can sleep without worrying about whether you have chosen the right loosy encoder.

You no longer have to worry about
- whether a new and improved version of your encoder of choice comes along, or
- somebody reports a problem with your encoder of choise, or
- your encoder of choice faires badly in listening tests.

You will be safe and have the power to automatically reconvert your entire collection to your encoder of choice.

However first and foremost it is your choice and what is important is what your gut feeling teels you - or dare yourself with your personal ABX tests to compare different encoders and settings.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #15
Quote
Even at 320 kbps you might be able to find tracks where you feel a difference between CD and 320 kbps on your type of equipment.


I know, lossy encoders will always have a problem with certain things, regardless of bitrate...but the odds are reduced a bit.  I did have a really good example ogg from a george carlin skit that was ABX'able easily even though it's bitrate was approaching 400 kbps at times.  Wish I didn't delete it.  The white noise from the mike was changing amplitude drastically between voice and silent parts.  May of been an old codec issue or something...enough of that though.

Quote
A problem like bad distortion on 3rd generation iPod turns up.


That kinda blows, I assume they'll have a firmware or something to fix this.  I hope anyway.  Not that I own an iPod...but was considering one.

As for lossless... I agree.  But there are a couple of reasons I didn't:

1) Was considering buying the aforementioned ipod to integrate into my car stereo, hence the choice of AAC as my encoding medium.  The size of the library I want precludes using WAV, so I substituted the highest AAC compression.

I am aware kenwood makes a flac compatible jukebox, but it's expensive and kenwood is not on my A list of headunits to purchase.  I have a hard time justifying dropping extra cash on a screen with dolphins and spaceships flying around.

2) Home computer did not (and still doesn't) have sufficient hard drive space to encode to flac AND maintain a complete backup of the library (~50 gb x 2).  Flac takes up almost 3x the space as the AAC's.  Also want to be able to syncornize between the car and the computer without keeping a separate library (FLAC and AAC and yet another harddrive)

3) At the bitrate I'm using the codec can do very badly in future listening tests and it won't matter any since the bitrate is overkill in the first place.  Or at least that was my thinking before the most recent listening tests concluded.

As for the trust in codecs bit I wasn't referring to a particular version of a codec as I was the user behind it.  The internet has no shortage of extremely poorly encoded 192 kbps mp3's from people who have little knowlege of what they are doing or are holding onto obsolete rationalle (i.e. true stereo w/ lame).  I've even come across some horrible 320 kbps mp3's...using a first gen blade encoder or something.

An addtional reason for choosing the transcoding preset on the AAC's I didn't mention in the previous post was also so that I could encode full bandwidth.  (Lame lowpasses at 16 khz to elminate certain artifacts, and make better use of available bits).  I know I can hear to 18-19 khz... so I must be missing something, whether it's easy to discern or not.  The same goes with frequencies below 20Hz...98% of people cannot hear or reproduce these, but the those who can, the effect is definetly noticeable.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #16
On an aside, someone needs to dig out an old thread I recall where 320kbps was better than --APS on a fair number of samples... I think that it was one of the catalyst's for -Z being intrigated, IIRC. 

Two other points: I cant believe modern portables have problems with 320kbps. And Xing at 256kbps, as a half-way-house to other encoders at 320kbps, is comical.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #17
this topic is a far from original post but... my understanding is that --preset insane is currently the highest quality u can get with lame (and probably the mp3 format). if u want to use to use mp3 (for a reason or another) and if u want to be sure of your encoding (some talk about "safety"), why not going for insane? it is still much smaller in file size than any lossless encoder around and arguably overkill!

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #18
Quote
this topic is a far from original post but... my understanding is that --preset insane is currently the highest quality u can get with lame (and probably the mp3 format).

not that I would realy use it, but what about 'freeform' bitrates above 320?! ...
are they very very experimental, or do THEY offer the highest (though unwise) quality?
Nothing but a Heartache - Since I found my Baby ;)

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #19
Quote
Lyx:
It's not like the community is infalliable either...remember when r3mix was 'transparent'?...remeber when joint stereo was 'bad'?.

Not allowing people to deviate slightly from the suggested norm is not a good idea.  Consider this scenario...what if instead of following the good advice of people on this forum I decided to follow what the media was telling me (via ads, articles in magazines, bestbuy salesmen)?  All of a sudden our point of reference shifts...now a 112 kbps WMA is 'CD-Quality' and 96 kbps is good enough for joe average listener and 128 kbps is for audiophiles and 192 kbps is overkill.  This is certainly what microsoft would have us believe...or sony with their ATRAC...and Apple with their 128kbps AAC ect...).


All of your above examples are cases which are probably not based on blind-listening tests(especially r3mix). And for sure not based on such demanding listening tests as they happen at ha.org.

Quote
Why can't I demand a greater margin of safety in my encoding?  Even if it is entirely for 'piece of mind'.  Especially when I browse over to the 3.96 vs. 3.92 thread and find that there is possible degredation in these presets you refer to previously as though they are absolute and irrefutable.  I just do not have the time or patience to ABX everything I encode just to be sure that APE is sufficient...


Read my explanation on under which circumstances such listening tests happen again. Those listening tests are already extreme-cases tested with "unrealistic" methods (the "training-phase", which doesn't happen with normal listening).

What i'm trying to explain: APE is already overkill and has a large "extra-safety". The difference between APE and API is just that APE at least uses up the bits in a reasonable way (VBR).

You're right, i don't have to mind if others choose to encode to API. Its not my music. And yes, the only downside is increased filesizes.

But that doesn't change the facts:
- lossy codecs are about getting the best quality/filesize ratio by dropping the information which cannot be perceived
- API is "insane" and inefficient concerning the above - thus, its inefficient for lossy purposes
- API does not make the audio technically the same as the original

conclusion: API is inefficient for lossy and lossless purposes

Off course someone can freely choose to encode audio in an inefficient way - to each his/her own - i don't have the right to enforce anything on someone else. I'm just telling the facts + a logical conclusion.

- Lyx

edit: what you basically want is a lossless codec which has the bitrates of a lossy codec - and your conclusion how to archive that is going full-bitrate with a lossy codec. The problem with that thought is that a lossy codec stays lossy no matter how high you pump up the bitrate.

Quote
this is not a case of yes/no, black/white, bush/gore.

There is a 3rd possibility: hybrid-codecs...... problem: no hardware support.
I am arrogant and I can afford it because I deliver.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #20
Quote
Two other points: I cant believe modern portables have problems with 320kbps.

Well, the iPod have been reported to have problems. And I know for sure that Philips DVD-players does not support 320 kbps at all. So there is definitely modern hardware out there that have such problems.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #21
Quote
Lev:
I cant believe modern portables have problems with 320kbps


Further information in 3rd generation iPod available here:
iPod badly distorted

Also experienced problems with jitter on low cost mp3 flash players at 320 kbps.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #22
Quote
QUOTE (holkie @ Jun 16 2004, 10:24 AM)
this topic is a far from original post but... my understanding is that --preset insane is currently the highest quality u can get with lame (and probably the mp3 format).

not that I would realy use it, but what about 'freeform' bitrates above 320?! ...
are they very very experimental, or do THEY offer the highest (though unwise) quality?


good point actually. i was referring to highest recommended settings with lame. i wonder if extra bitrate above 320k would improve encoding results but what i understand is that it makes a file incompatible with any portable mp3 player...

about those ipod problems i havent encounter any with ipod from 1 and 2nd genrations (mousewheel ones?).

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #23
Quote
On an aside, someone needs to dig out an old thread I recall where 320kbps was better than --APS on a fair number of samples... I think that it was one of the catalyst's for -Z being intrigated, IIRC. 

I'm to blame for that one...

http://www.hydrogenaudio.org/forums/index.php?showtopic=7783

...though you can "fix" any problem by throwing more bits at it.

Cheers,
David.

LAME @ 320 Vs. Other mp3 Encoders @ 320

Reply #24
There may be just one or two reasons to favour 320 CBR (Alt preset insane) over Extreme (a VBR preset).

1. You don't trust VBR - a valid point at one time, but with the presets, VBR WORKS!
2. If LAME behaves like it used to, encoding at high CBR bitrate tends to be fastest

As for quality, if a sample is killer for Extreme, the only way Insane would rescue it, is if it was a problem relating to VBR.

CBR 320 will be throwing bits at places where they aren't needed.

 
SimplePortal 1.0.0 RC1 © 2008-2019