Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: LAME -V3 vs -V4? (Read 6808 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

LAME -V3 vs -V4?

Hello,
after reading here in the forums, I think -V2 might be transparent enough. But what about -V3 vs-V4 vs -V5? The wiki only says:
Quote
[...] while -V 4 should be close to perceptual transparency.
I'm about to encode my FLAC 24/96* classical music collection to mp3 (should be good enough, no need for Opus or something else at same bitrates, right?), but need to decide whether to go with -V2, -V3, -V4, or maybe even -V5?

* After I have learned that 16/44.1 is all it really needs.

Re: LAME -V3 vs -V4?

Reply #1
after reading here in the forums, I think -V2 might be transparent enough. But what about -V3 vs-V4 vs -V5?
There is only one person that can determine whether any setting is transparent for you, and that is you yourself.

I'd say pick a few music tracks from your collection, convert them to V5 and perform an ABX test. Are all tracks transparant, then you're done, V5 is your setting, or you can try V6. If not, do the same thing with V4 etc.

Whether a setting is transparent depends on your taste in music, your ears, your listening equipment, whether you are hung-over or well-rested etc.
Music: sounds arranged such that they construct feelings.

Re: LAME -V3 vs -V4?

Reply #2
Quote
There is only one person that can determine whether any setting is transparent for you, and that is you yourself.
I have also read this template reply so that's why I had to ask, because it's not reliable enough as you yourself say:
Quote
Whether a setting is transparent depends on your taste in music, your ears, your listening equipment, whether you are hung-over or well-rested etc.

Maybe time to remove this template reply? ;)

Re: LAME -V3 vs -V4?

Reply #3
because it's not reliable enough as you yourself say
That's the most reliable way. If it's not enough for you then use lossless.

Re: LAME -V3 vs -V4?

Reply #4
Since you're in charge of running your own ABX tests, you can easily improve the reliability of the results by only conducting tests when you're feeling alert. If you don't want to run your own tests, just use the recommended settings. Those are really the only two options. Nobody can tell you if V2 is overkill except you.

Re: LAME -V3 vs -V4?

Reply #5
If you're not concerned about file size you can go with a "higher quality" setting, even V0 or 360kbps CBR.      But, we can't really say higher quality if we can't hear a difference.  

More compression (smaller files) is the only reason to choose a higher V-number.

Re: LAME -V3 vs -V4?

Reply #6
There is a reason not to choose a low "quality" even if one cannot hear the difference in testing: you test ten files, you convert a thousand, there is quite some chance that one of the 990 is a bit different.

The solution is simple: keep the originals. Then you haven't once and for all "chosen" the low bitrate. Consider the lossy files disposable. If one of the 990 files sound bad ... replace it. If enough of them do ... run the whole procedure at a higher bitrate. 

Digital music stores can have all sorts of reasons to choose 256/320 kbit/s - high numbers look good and therefore they sound good to customers who hear with their eyes - but one very good reason to go overkill is that you don't want to pay a big listening test panel to find a transparency level for each file.

 

Re: LAME -V3 vs -V4?

Reply #7
Those settings usually sound very good or even perfect.
Personally, I see it like this;
With todays storage no reason to go lower than V3 or V2.5 when
you want a smaller file size. 

For me  it takes -V1 ~ 224k to get a robust all round quality.  Also -V2 --vbr-old ~ 190k
works well for some reason. 

Given this, Its not a issue to go -V1 as its still lower filesize than the standard 256 / 320 .
At higher bitrates (200+) I am not sure the files are so 'disposable' . I think It can work even
as one archive.

Re: LAME -V3 vs -V4?

Reply #8
In short... use V5 (130kbps) and forget about it.

I tend to see V5 (130kbps average) as a pretty good default setting for most people, especially if sound quality/file size efficiency is of any real concern. besides, based on listening tests around here, V5 scores well enough and when just sitting back and enjoy your music on a typical listening device, which is typically going to be the general use case, it's not likely people are going to notice any sound quality hit. even if there is on occasion, it's small enough not to really matter in my estimations. because unless the audio quality drops low enough to where you can pretty steadily/easily notice a difference, it's generally going to be transparent a very high percentage of the time.

if I was going to use anything higher than V5 I would probably just go straight to V2 (190kbps average) and forget about it since I would guesstimate that will offer a high enough portion of sound quality that MP3 has to offer, but without totally sacrificing some level of efficiency, as I figure any higher than that, your probably better off switching to another format (Opus or AAC(AAC-LC)) to gain better efficiency.

with that said... I get many will argue with today's storage space (which tends to be a lot for minimal $) that one might figure why risk any sound quality drop for a file size gain that's nothing too large, and while I get the reasoning, I tend to be more of the mindset that when using lossy audio you want to keep file size efficient as after a certain point, like if storage space is not a issue, you might as well just use lossless if your paranoid about sound quality. that's kind of why I tend to side with V5 as it's pretty much the same as the higher bit rate settings in general and is much more storage space efficient.

p.s. I always suggest keeping the original FLAC file(s) as on the chance you need to convert to another format you can always do this easily enough in the future and it's always wise to keep the high quality source file around as a 'just-in-case' sort of thing. but the good thing about MP3 is it's pretty much guaranteed to work on anything that supports lossy audio, which is why it's always a safe choice and it's still pretty safe to say V5 (130kbps) is of high enough quality as if you play random songs at this bit rate, I would be surprised if random person listening to it would complain, since if there are any flaws they won't be obvious enough.
For music I suggest (using Foobar2000)... MP3 (LAME) @ V5 (130kbps). NOTE: using on AGPTEK-U3 as of Mar 18th 2021. I use 'fatsort' (on Linux) so MP3's are listed in proper order on AGPTEK-U3.

Re: LAME -V3 vs -V4?

Reply #9
Personally, I see it like this;
With todays storage no reason to go lower than V3 or V2.5 when
you want a smaller file size.

Unless, of course, one of your parents has a radio which supports MP3 on CDs but no USB support and you need to fit their favourite band's discography onto a single CD for them.


Re: LAME -V3 vs -V4?

Reply #11
And even then if you have aux or bluetooth support it could be argued you're better off using that than the cd player with some sort of smartphone or mp3 player.

Re: LAME -V3 vs -V4?

Reply #12
And even then if you have aux or bluetooth support it could be argued you're better off using that than the cd player with some sort of smartphone or mp3 player.

But 1) you don't, and 2) even if you did, good luck explaining either aux or bluetooth to your sufficiently technologically illiterate parents.

Re: LAME -V3 vs -V4? (Edit: It's going to be Opus now)

Reply #13
Thanks everyone.

In short... use V5 (130kbps) and forget about it.
[...]
This is probably not too far off where I'd have went with if I went with MP3. Something like -V4 or -V4.5, after I did more reading.


After further reading I think I'm going to use Opus after all, the savings in disk space are just too nice (yes, even today it matters for me), despite that the support for mp3 is probably still wider and more and more support is added for Opus.

I don't know if LAME still has issues/problems even with higher bitrates like -V3, and even -V2, but this is not a reason why I choose Opus. Many say on this forum and elsewhere that even at 80 kbps Opus sounds transparent to them (some say even below that) and 96 kbps is good enough for most. What I get from reading the listening tests and forum replies in general is that Opus at 128 kbps sounds transparent to almost everyone, like 99.x %. I don't think I'm going to use 128 kbps, maybe 96 kbps, not sure if 112 kbps are required (have many Harpsichord tracks, so maybe it is, probably isn't tho. A few quick tests will be enough for me to find out and remove any doubts). Also read on here that Opus requires only about 60 % of MP3's disk space (imagine 1TB (MP3) vs 600GB (Opus) -> 400 GB saved). Say I'd went with MP3 at 160 kbps for being maybe/probably more than enough, so with Opus that would be 160 kbps*0.6 = 96 kbps.

I did a quick test:
Code: [Select]
$ sudo pacman -S opus-tools
$ opusenc --bitrate 112 01\ -\ Allegro.wav 112--01\ -\ Allegro.opus
$ opusenc --bitrate 96 01\ -\ Allegro.wav 96--01\ -\ Allegro.opus

and I can't hear a difference between wav; LAME's V2, V3, V4 and Opus. V5 not tested but it's going to be Opus anyway.

PS: I also have a lot of electronic dance music.
PPS: Too bad can't edit the main title to "LAME -V3 vs -V4? (Edit: It's going to be Opus now)".

My hardware SCHIIT Modi 2 DAC + YAMAHA MSP7 speakers. (not a balanced DAC but no additional noise after testing by unplugging XLRs/same outlet, so all good). Headphones are better for testing but my HD600 are not here currently (amp would be SCHIIT Magni 3+) and I prefer speakers by now anyway.

Re: LAME -V3 vs -V4?

Reply #14
Go for V4.  It used to be my pick for old portable players. Overall 150k vbr
and quality benefits over V5 in some cases. Its a notch of headroom
over v5 and 128k listening tests.  

Re: LAME -V3 vs -V4?

Reply #15
In short... use V5 (130kbps) and forget about it.
[...]
After further reading I think I'm going to use Opus after all, the savings in disk space are just too nice (yes, even today it matters for me), despite that the support for mp3 is probably still wider and more and more support is added for Opus.

Many say on this forum and elsewhere that even at 80 kbps Opus sounds transparent to them (some say even below that) and 96 kbps is good enough for most.

If you think V5 (130kbps) on MP3 is good enough... if your using Opus I suggest seriously considering 64kbps as that's probably still 'good enough' for many people as if your going to use any higher than that, I suggest just going straight to 96kbps and forget about it. speaking for myself... I would likely struggle to notice a difference on a typical Opus 64kbps music file and I highly doubt I would notice any difference at 96kbps (if I could, I suspect it would be rare/small enough not to matter).

bottom line for me when it comes to Opus... use 96kbps or less otherwise your better off using a more standard format like MP3 overall since that's Opus main strength is getting good enough sound quality at bit rates that would be too low for MP3 (so you can roughly halve the storage space with a sound quality that's plausibly very similar overall). basically once you start getting into around that 128kbps+ range, while Opus is a strong performer, MP3/AAC(AAC-LC) starts to look more appealing overall given they are more widely supported.

or another way I might put things, here is my suggested minimums for each of the following formats...

-MP3 = V5 (130kbps)
-AAC (AAC-LC) = 96kbps (I suggest FhG or Apple (which I do through Foobar2000))
-Opus = 64kbps (while one may get away with going lower, I kind of figure this is somewhat the sweet spot since it's probably not worth going to say 48kbps and saving 16kbps(or 25% less storage space) for potential further drops in sound quality)

but come to think of it... while I mentioned already that once you get into the general 128kbps or so range that I would probably just stick with MP3 given the universal support, and while I mentioned Opus @ 96kbps is a solid option worth considering, I kind of figure once you hit 96kbps you might be better off with AAC(AAC-LC) given it's the next best inline with being supported behind MP3 and I doubt many would complain about it's all-around sound quality. hence, that's kind of why I figure if someone is using Opus with music that 64kbps is probably the sweet spot since it gives a legitimate storage space savings over MP3/AAC without sacrificing sound quality too much.

p.s. on Opus for speech, I don't mind using as low as 13kbps as that sounds surprisingly good for speech considering the very low bit rate. because while it's not transparent, as you can tell the difference between the original source and the 13kbps Opus file, it's still 'good enough' considering the major storage space savings and with speech I am not as picky with sound quality as I would be with music.
For music I suggest (using Foobar2000)... MP3 (LAME) @ V5 (130kbps). NOTE: using on AGPTEK-U3 as of Mar 18th 2021. I use 'fatsort' (on Linux) so MP3's are listed in proper order on AGPTEK-U3.

Re: LAME -V3 vs -V4?

Reply #16

But 1) you don't, and 2) even if you did, good luck explaining either aux or bluetooth to your sufficiently technologically illiterate parents.
But mp3 cd support on a car stereo tends to have a really terrible UI experience?

EDIT: I'm just saying that playing a mp3 cd on a stock head unit in a car potentially tends to not be particularly accessible to the technologically illiterate even if someone else makes the CD.

Re: LAME -V3 vs -V4?

Reply #17
- SCHIIT Modi 2
- SCHIIT Magni 3+
- YAMAHA MSP7
- HD600
> Lossy 96 kbit/s music

gold plated toslink fan

Re: LAME -V3 vs -V4?

Reply #18
- SCHIIT Modi 2
- SCHIIT Magni 3+
- YAMAHA MSP7
- HD600
> Lossy 96 kbit/s music

Well, if it sounds transparent or satisfactory to them, then what's the problem?

Also, "lossy 96 kbit/s music" indirectly implies that all lossy formats are the same and equally good(/bad) at 96 kb/s.

Re: LAME -V3 vs -V4?

Reply #19

But 1) you don't, and 2) even if you did, good luck explaining either aux or bluetooth to your sufficiently technologically illiterate parents.
But mp3 cd support on a car stereo tends to have a really terrible UI experience?

EDIT: I'm just saying that playing a mp3 cd on a stock head unit in a car potentially tends to not be particularly accessible to the technologically illiterate even if someone else makes the CD.

I've had a couple car stereos with mp3 CD support.  In my opinion, it was then, and even now, better than trying to control the music over Bluetooth.  I find Bluetooth slow to connect, laggy, and both devices for both audio and controls will default to the lowest common supported format. 

Re: LAME -V3 vs -V4?

Reply #20
Hello again, I've done more reading and testing. I'll make it short(er) this time.

--bitrate 64: Could notice it [vs original], but depends on style of music (I noticed it much less on EDM, than classical music, at least so far). I'd use it for, say, game/voice chat and such, not high quality music streaming, at least not where the user pays for the service. I read that when SoundCloud switched to Opus @64 many/most users didn't notice the change. Still I hope they, or any other such service, offer higher bitrate at least for those who pay, as Opus@64 is simply not transparent enough.

--bitrate 80: Sounds transparent so far, but I did much fewer tests vs 64. I'll do more.
To be on the save side I may use just use 96. The only question is if --bitrate 90 is enough too (I get it's not a small difference, but one could fit 6,7% more tracks that way). I guess, after all, only my ears on my music collection can answer it.

[...] I suggest just going straight to 96kbps and forget about it. [...]
After all the reading and testing I agree with this. I would probably not go above --bitrate 96, even for archiving, at least so far.

Regarding the SCHIIT name, yes, many jokes can be made :) But they, at least for one product I have, make it justice: There's a topic called "Please help! Schiit Modi 2 not being recognized anymore by Windows?". I confirm this problem on my Modi 2 (I have to warm up the chip to make the DAC getting recognized, just as suggested there). I've read they use the affected CM6631 on some of their other products as well.

I'm not going to dig-up this again when I finally decided which bitrate I'll use. Feel free to post something notable.

Re: LAME -V3 vs -V4?

Reply #21
--bitrate 64: Could notice it [vs original], but depends on style of music (I noticed it much less on EDM, than classical music, at least so far). I'd use it for, say, game/voice chat and such, not high quality music streaming, at least not where the user pays for the service. I read that when SoundCloud switched to Opus @64 many/most users didn't notice the change. Still I hope they, or any other such service, offer higher bitrate at least for those who pay, as Opus@64 is simply not transparent enough.

--bitrate 80: Sounds transparent so far, but I did much fewer tests vs 64. I'll do more.
To be on the save side I may use just use 96. The only question is if --bitrate 90 is enough too (I get it's not a small difference, but one could fit 6,7% more tracks that way). I guess, after all, only my ears on my music collection can answer it.

I would probably not go above --bitrate 96, even for archiving, at least so far.

Since your on the topic of Opus...

Yeah, I am sure some can notice 64kbps (I don't doubt this) but if your not comparing to the original lossless file and just sit back and enjoy your music I suspect it's not as easy to notice which is why I am not surprised many can't notice any obvious differences (like you mentioned with the SoundCloud comment). but for voice I think 64kbps is probably a bit too high, although I guess it depends on how picky one is with voice quality. because personally, off the top of my head, I would probably not use any higher than around 32kbps for Opus on speech (maybe 64kbps on the high end), although I usually just default to 13kbps (although I would avoid going any lower) since, while noticeably worse, is still good enough as it does not sound significantly worse than a higher quality source and the file size is drastically smaller. so, like music, I think in terms of speech with Opus it would start to lose it's appeal over AAC/MP3 if the bit rate is not more on the very low side of things since once the bit rate goes beyond a certain point a person would probably be better off using MP3 since it's so standardized.

but in terms of 80kbps... it was said around here that Opus @ 80kbps is similar to LAME(MP3) @ V5 (130kbps). but anyways, while I am sure 80kbps is safer than 64kbps in terms of sound quality, to me there is not enough file size savings with 80kbps over 96kbps which is why I am of the mindset use either 64kbps or 96kbps when it comes to Opus and forget about it since those two settings pretty much cover the best of both worlds... like a very efficient file size without too much hit to all-around sound quality, and the other offers strong sound quality but still pretty strong in efficient storage space. although with that said... I am sure a person could still make a solid case in favor of 80kbps (or the like) for someone trying to 'fine-tune' things to that edge of barely any noticeable sound quality loss but as minimal file size as one can go etc.
For music I suggest (using Foobar2000)... MP3 (LAME) @ V5 (130kbps). NOTE: using on AGPTEK-U3 as of Mar 18th 2021. I use 'fatsort' (on Linux) so MP3's are listed in proper order on AGPTEK-U3.