Hi. I have some possibly general questions about encoding ... but i need some deep answers

Ok .. here is the thing. I like to do the encoding myself because, in general, all the mp3s [L.E. legal music] i've downloaded form the p2p networks are encoded badly. They have 200 Kb/s + and they sound horrible.
So ... now i'm encoding from some lossless format to some lossy one (like mp3).

Hm ... i need to know if i can improve the sound quality in my encoded mp3s. I'm using this setting >> -b 128 -m j -h -V 2 -B 320 -q 0 . This means that i'm using a rage from 128 Kb/s to 320 Kb/s , encoding based on quality ... variably bitrate ... (quality 0) V 2 ... This gives me, in general, about 205 Kb/s mp3 files.
I'm using RazorLame to set the encoding. It doesn't have many options ... so i'm wondering >> what i can do to increase quality for ...let's say 210 Kb/s average ?

What program do you recommend for encoding ... with options for ... let's say advanced users ?

One more q ... doesn't ogg produce a better sound for about 200 Kb/s ? I was reading this topic... i realy don't know if it is subjective or professional ... but....what do you say ? >> http://www.digit-life.com/articles/oggvslame/

Hi,

first of all there is something I must say about downloaded files: if you download from p2p you can't be sure about the source of the files. ~200Kbit/s is a high bitrate, which is trasparent to almost everybody (please see hydrogenaudio Wiki or FAQs for clarifications), the files you've downloaded may have been transcoded, who knows?

Today, the best encoder available is LAME. Current version is 3.97 and you can download the executable from rarewares.org. One interesting thing about LAME is that it uses presets: these are pre-configured modes that give the best quality in a given bitrate range. If you use --preset standard (now called also -V 2) you'll get the best settigns for the 170-210kbit/s range. If you're worried about quality you can use -V 0, but please notice that for almost everybody it will be exactly the same as -V 0

So I reccomend you to use lame in this way:

Lame.exe -V 2 --vbr-new in.wav out.mp3

please see this post for more details

Remember that you can always use a fronted for lame, to rip your CD's and convert them directly, in this case I reccomend you Exact Audio Copy

IIRC, almost everybody here uses -V0, -V1, -V2 or -V3. Including the experts. The other options are believed to override what the -V settings use, and have a high risk of making the sound quality worse, or increasing the bitrate without improving quality.

So I'd say just use whatever -V setting gives you the bitrate you want.

For MP3 most people here use LAME. The notes on it are here:

http://wiki.hydrogenaudio.org/index.php?title=LAME

I think what is pretty much the standard setting now is:

-V 2 --vbr-new --preset fast standard

which, as it says, generally gives bitrates something around 170--210.

Many windows users seem to be ripping with EAC. Here's a page on using the two together:

http://wiki.hydrogenaudio.org/index.php?title=EAC_and_Lame




Not by enough to be noticeable - or not by almost all listeners. If you were talking about bitrates of 48kbps, perhaps. By 128kbps most modern codecs are more-or-less tied.

Is there really any need to discuss about illegally gained material from P2P networks here? To me "some" lossless format sounds like you actually didn't encode these source files by yourself.

There is a last thing you must remember: if you use iPod you could have some skipping problems with -V 2. With iPod I suggest you to use fixed bitrate, I use 192kbit/s which is more than good for portable use

Lame.exe -b 192 in.wav out.mp3

Please don't draw conclusions about sound quality from reading this article. Not only it's horribly outdated but additionally uses non-scientific methods to assess the quality of the codecs, therefore it's practically useless. If you want to know what codec performs better, ABX yourself or rely on various listening tests performed here on the forums.

I'm sure they talk about legal music here (which you can - surprisingly - also get on P2P networks)... right, guys? You wouldn't risk a warning or anything, would you.

Raiden i know i can find all i want on this forum. It is simply amazing how much i can learn from it ... but i really can't read in a small amount of time ... ALL this. Hm... i probably need a couple of days to analyze and understand the majority of the topics. After that ... i can say that i'm a PRO

Junon This was an example. Of course ... if you want i can give you some radio stations (and not mention the word "p2p") that stream at 192 Kb/s in mp3 format ... encoded very bad. They are perfectly legal you know ...

giopiar I have a mp3 player but it is not ipod. Ipod ... not my style.

Nick E, Axon, giopiar i will remember this. So ... it is good after all to use the default presets. I thought that are only ... for noobs . Sorry

10x all for reply. I will check this topic, maybe something new appears.

P.S. CiTay P2p networks can and are used to transfer legal content. It can be video, audio or text. For example ... video > trailers < ... music > free mp3's to promote some singers or albums < . Of course we are talking about free mp3s encoded bad

Only on older models... Lame VBR works great on 5th gen for sure, 4th as well i believe...

@ bogdann

OGG Vorbis is indeed a format to check out. most OGG users in HA agree transparency is achieved at -q5 (around ~160 kbps). for MP3, most LAME users in HA agree transparency is achieved at -V2 (around ~190 kbps)... please see HA wiki for both LAME and OGG Vorbis in HA

you be the judge.

I know the wiki recommends V2 but i think you should try ABXing the different settings on your portable. I tried it, and i couldn't even ABX -V6. So I use V5 just to be safe. You could save a lot of space by doing a simple ABX test.

This is just it, isn't it? My understanding, possibly wrong, is that that setting

-V 2 --vbr-new --preset fast standard

is the LAME people's recommendation rather than just the wiki's. But, of couse, that doesn't mean that just because they consider that to be satisfactory many, even most, listeners will find not lower bitrates just as acceptable.

As it happens, I don't use LAME any more myself. I use the Core Audio AAC encoder in Mac OS X. But here's one of the most widely seen listening tests around. It's at 128kbps:

http://www.listening-tests.info/mf-128-1/results.htm

One encoder is significantly worse. There is no significant difference between the performance of any of the others, and the conclusion is that "with the current encoders, the quality at 128 kbps is very good" and it's mostly no longer worth doing tests at that bitrate.

Ogg's still better in that test

Seriously though, is it really satisfactory for all the people in this topic to say basically "I don't think there's much difference, therefore let's use lower bitrates', or 'I did a couple of listening tests therefore X..'. Doesn't sound very scientific (almost against TOS, even!).


There's a good comparison at 160 kbps on this forum from around the same time as that test.

Also, isn't that test also out of date (Dec 2005)? I thought LAME got a lot better since, and presumably aoTuv's got better also.


The former 'codec wars' seems to pretty much died off, indeed. But my problem with all this is, there seems very little scientific evidence showing well, anything, and the majority of 'research' seems to have been done by people at home with out telling us much except they gave a casual listen or casual ABX.


I just wrote a post at my forusm explaining why I use Ogg Vorbis, and the reasons are probably out of date. I'll give thought to including LAME MP3 as well. But I wonder if we will see any more good quality tests, or if we'll be relegated to 'whatever, it all sounds the same' thinking that I thought would have been gotten rid of, at least at these forums.

BTW: What's wrong with this article?
http://www.digit-life.com/articles/oggvslame

I'm not suggesting I have a problem with what's been said, I'm just curious as to the science/logic that's outdated (I'm a bit out of the loop).


- Spike

Just use -V2 --vbr-new.

-V2 --vbr-new and --preset fast standard are the same thing (--preset fast standard is outdated).

In 3.98 my understanding is that you will just need -V2, as the newer VBR routines will be used by default.

When i did the test to get a rough estimate of what bitrate to use, i tested on music that i listen to-- not problem samples; because the point of the test was not to see if i can hear the artifacts, but to check how low i can go before i can notice a difference.

Since the scope of the tests are so different, i don't think personal tests are less valid.

Wow im impressed that people took time to answer a question like this (asked 10.000 times before) violating the rules regarding claims of compressed files sound quality without proof and implying download of illegal files. (come on guys, you think he is going to Jamendo?)

sizetwo My primary question was if i can improve those settings for Lame. Be sure that i've spend a little time searching the forum and i saw only settings like V 2 ... i was wondering if all the people use those settings or are written there only to keep things simple.
Second q was about the mp3 and ogg encoding. I know that ogg is better at low rates but i didn't know for sure how high you must go with the bitrate for transparency.
Lame has options for the quality of encoding. Very fast encoding ... poor quality of sound. Slow encoding ... processing the file deeply ... you get a high quality file. This is a thing that all users know about it ... it needs no proof.

Pff... nobody said anything about this. Here ... let me search the google for you ... here you go: the first result was this > http://en.wikipedia.org/wiki/Peer-to-peer . You probably know about wikipedia ... it is not illegal.
This is a quote from the site mentioned above >>
So i repeat ... i'm talking about legal content.

L.E. >> I've set the --vbr-new method for this setting >> -b 128 -m j -h -V 2 -B 320 -q 0 and it has huge improvements for speed. It is about 3 to 4 times faster to encode in vbr-new rather than the old algorithm

Well "-h" and "-q2" are the same ... So don't use "-q0" which is useless in this case ...

Did you bother to read this thread? FORGET SWITCHES! Presets are already tuned for the best quality, and you risk to reduce it!

Yes, presets have been improved along the years, and it is safe to use it. LAME has the ability to use preset in order to ensure that the user won't be bored about complicated switchs related to quality. Switches should be used in extremely rare cases, for exemple for compatibility. Juste use "-V2 --vbr-new" instead of this non-sense switch...

even after several years after the presets have been developed, people still think adding custom switches will improve quality. i hope future lame versions will disable them altogether, or at least have a message saying something like: your switches will only degrade quality, continue? (Y/N)

There is no proof because it is completely wrong.

Ogg Vorbis is better than mp3 at the same bitrate but the Lancer builds encode 3 times faster than the fastest LAME 3.97 builds (on my com).

Just use -V2 --vbr-new. Unless you're a LAME developer or codec tester, you'll be deemed a noob if you use custom switches.

Vorbis (Ogg is the name of the container format) is better than MP3 at low bitrates and high bitrates. It is better than MP3 at any equal bitrate.

See this page: http://wiki.hydrogenaudio.org/index.php?ti...nded_Ogg_Vorbis

Scroll down, see the -q table and note number 4 under the table:



Add one more (i.e. -q 6), and 99.9% of the population will, I believe, say that transparency is achieved. Even some die-hard paranoids in HA don't go higher than -q 7, which roughly corresponds to 224 kbps.

Most of us mere mortals already find -q 4 to be transparent, i.e. approx. 128 kbps.

Please, before doubting anything, or even opinionating, do check HA's wiki.

Lol at this post:


Quote of the week, I love it.


Why make things more complicated than they have to be.

Kanak, I don't consider your tests invalid, I'm just saying, for something so contentious (in my view) it seems a little risky to make bold claims you can't back up.
Although I'm guilty of that myself sometimes.



Question: Is "-V2 --vbr-new" going to give me VBR MP3 at the highest quality, varying between 128 and 320 kbps where necessary, with minimal artifact?

That seems to be kind of missed, with all this talk of using some standard version, noone's actually talking about the quality/desirability of options, just that they're 'standards'. I still feel slightly confused about which preset to use!

Regards,
- Spike

I'm just thinking about the 128kbps multiformat test here, but that difference was negligable in that test - all encoders tied - although Vorbis did score marginally better than LAME. A few people may be able to tell the difference between LAME -V1/-V2 and Vorbis -q7/-q6 on a few tracks. I think such bold statements need to have some context.

-V2 --vbr-new would have no hard floor, so it will go below 128kbps, often. That's good, when it's not necessary to use more bits.

Whether -V2 is desirable is a question only you can answer. I use -V5 and find it absolutely fine. Encode a few tracks using a few of the presets and see what sounds good to you.

I've read about this ... but i thought that -q0 overrides the -h setting. I forgot to take out the -h.

About the presets ... i think the presets are tuned very good BUT ... if you read what are the default options you will see that some are marked "recommended" or default. So in adition to the presets you may use switches ... you can enable some options or disable them ... but the base remains the preset. This is a matter of preference

P.S. Pls don't argue about the preset vs switches thing ... it makes me feel bad because i never intended to go in this direction.

At least once a week this same question arises AGAIN...

The wiki must be made more prominent since its article on LAME presets is really good:
http://wiki.hydrogenaudio.org/index.php?ti...short_answer.29

Mantra: NO switches, JUST presets. They're there for a reason.

(Not) sorry to be blunt here, but if you are not willing to put in the effort to help yourself, then why should others put in the time to help someone who is too lazy to help him/herself? This is more than just parasitism - it is abuse.

http://www.hydrogenaudio.org/forums/index.php?showtopic=7333


No. It is a matter of comfort vs. education vs. ignorance. If you dont know about the inner workings of lame, then you have three options:

- the comfortable way: either use the presets or the recommendations of people who do know how lame works.

- the educated way: spent LOTS of effort learning how lames internals work and what kind of DBTs were done in the past. This will take months or even years.

- the ignorant way: play with stuff without having a clue what you are doing, decrease soundquality and then let placebo do its work.

- Lyx

You ask for recommended LAME settings and yet you don't want to hear recommendations?

OK, then some clarification: The presets are not some base from which you go on with switches - the presets are the optimum. By applying switches - which only makes sense if you encounter compatibility problems otherwise - you deteriorate your sound quality or unnecessarily increase file size, maybe even both.

another suggestion to any lame dev reading the thread: it would be great if in the documentation included with lame it is stated that usage of switches can only degrade quality and must only be used for testing or compatibility.

I am agreeing more with Joel Spolsky:

http://www.joelonsoftware.com/items/2006/11/21.html

and

http://www.joelonsoftware.com/uibook/chapt...0000000059.html

One more great article by joel, i agree. I wouldnt go as far as him, reducing it to just one option (i would keep "power off"), but overally, i fully agree - including the implicit suggestion regarding lame.

- Lyx

I agree !