Hello All:
Would love some information on VBR and ABR MP3's. When I started My Digital Music Collection a few years ago I started with 256 and 320 CBR MP3's. I usually ripped my music with WMP. Last year, I got rid of WMP and switched over to another media player. It was some free player which was called fre:ac which had a newer version of lame. I still used CBR's and I found them to sound better then the songs ripped by my old version of WMP, but I started messing around with other formats. I found that wma's (non protected) at 256 to 320 outshined my CBR MP3's at the same bitrate, so I started using the WMA's. This led to a problem. Even though my three mp3 players all suport wma's at 320, the wma's hit the battery hard. The wma's seem to suck about 50% more battery life then the MP3's did. (and I still had mp3 on my player, but majority of my music was wma.) So I did some research and found that VBR or ABR mp3's are the way to go, giving you better quality. I tried this and I really like the way they sound. I also downloaded media monkey and use this to rip my CD's. MM was a highly recommended media player and even though I dont know what version of lame it is using, it seems to do the job fairly well. So here comes my questions and sorry about the long introduction. I need some advice for the settings. I have played around with many and I have found that using the ~245 setting with no advanced settings seem to do the trick. I listen to metal, classic rock+pop/oldies, electroinica and jazz and I usually end up with a bit rate of 280 for metal and electronica and 255 for jazz and classic rock. I would love some advice some of my other settings and would love some opinions of my settings. Hopefully you will have some fun with this as I know some of you are audiophiles and would love your opinions. I am trying to get the best quality of my music with the lowest Bit rate possible so the battery doesnt get eaten up. Here are my settings:
1.~245 (no advance settings) This seems pretty good as stated before, but without any setting to let you know what you set the lowest bitrate for, sometimes I have songs that drop down to an "average bit rate" of 240. This happened with John Coltrane and Miles Davis. Also with Traffic and Steely Dan the average was about 260. I thought this was a bit low for those types of music. I thought Jazz and Progressive Classic Rock such as Steely Dan,Yes, and Pink Floyd demanded a higher bit rate and even though the music would peak up in the 320's at times (hence the advantage of VBR) this seems kinda low for an average.
2.~245 with advance settings of lowest bit rate of 256 and highest of 320. I have bit resoviour turned off. This setting is good as I get a higher avreage bitrate, usually 295-305. Pretty Close to the 320 mark I am sure that its just the same. Most all types of Music come out at around that Bit-rate. Big question about this is that if a song hits a passage of lets say 160, does the encoder still go that low since I set lowest to 256.?
3. 300ABR-I use this when I get a file that has an average Bit rate that I think is to low with the other previous settings. I heard its almost like VBR, but I cant seem to find anything on its actual behaviors. With this I get a a average Bit rate of 290.
Are the Bitrates I am choosing Overkill? What setting would you choose? I think that there might be a slight difference between a file playing at an average of 225 then 305 but it just might be like a placebo effect and once I know I am playing a song at a lower average bitrate it doesnt sound as good so my ears are helping me in my test..lol Does anyone hear a difference with large numbers of values in bitrates? I understand this is really difficult because so many people have different views. I know some people that say 192cbr is cd quality (i disagree, but if thats what your ears tell you then go for it) I just want to see what the majority has to say of where they like their average bit rate to be or what they think is needed to achieve highest quality. Love to hear from as many as possible
Hi and welcome.
I'm seeing quite a few statements about sound quality but no effort to communicate how you went about determining them.
http://www.hydrogenaudio.org/forums/index....showtopic=16295 (http://www.hydrogenaudio.org/forums/index.php?showtopic=16295)
The rules of this forum require this.
http://www.hydrogenaudio.org/forums/index.php?showtopic=3974 (http://www.hydrogenaudio.org/forums/index.php?showtopic=3974)
Hi and welcome.
I'm seeing quite a few statements about sound quality but no effort to communicate how you went about determining them.
http://www.hydrogenaudio.org/forums/index....showtopic=16295 (http://www.hydrogenaudio.org/forums/index.php?showtopic=16295)
The rules of this forum require this.
http://www.hydrogenaudio.org/forums/index.php?showtopic=3974 (http://www.hydrogenaudio.org/forums/index.php?showtopic=3974)
I'm not sure what you mean by determining quality. I read the rules of the forum and I am not understanding.I am so sorry I am new to this. Maybe these statements will help. My main goal was what experts and people think what the average bit rate should be or come to with certain types of music to achieve the highest quality (well the highest quailty that the MP3 can go, since MP3 is lossy) and smallest file size. Its hard for me to determine since I think my ears have a placebo effect. Its like once know what my MP3 is playing at, its like the one with the lower bit rate doesnt sound as good. I just dont know if this is because have become hard-headed and think a greater bit rate will be better even if that bit rate is only a difference of maybe 30 or 40kbps. I seached and seached this forum on these topic but I can't seem to find anyone that has asked such a spefic question about a VBR or ABR bitrate. Most of the post all refer to a CBR or a CBR vs a VBR but at what bitrate (nor have I read of anyone using ABR's in the 300bitrate range). If I hear from experts that have been dealing with VBR's for years (remember I am new to them) either by listening or by some sort of graph or tester (sorry i have no ideas what the names of these things are) This way I know what setting to choose, that I no longer have to worry about eating up to much space and perserving battery life with the best possible SQ. Hope this helps and that my posting is still allowed.
Its hard for me to determine since I think my ears have a placebo effect.
The first link I provided tells you how to avoid this very real and negatively influential phenomenon. Please read it.
In the second link please read about rule #8, in the first post as well as in the later post that describes the rule in more detail.
At this point I think you can safely discard any of the conclusions you've presented about different compression settings and how they affect sound quality.
ThomasG3rd:
- Your original posting contained a few subjective opinions of the type that this codec of that bitrate was better than (“outshine” is quite a strong word) some competition. As long as those statements refers to sound quality (rather than e.g. to battery life), then they are per the terms of service not welcome on this forum unless backed by evidence (see the TOS) or infeasible to test (like “I had to rebuild my house after the fire, but at least I got rid of that annoying boom at 67 Hz” would be a statement way beyond your ability to back up by a listening test). Of course moderators will likely not knock you for stating “I tried to save space by going mono, but that sounded too annoying”.
- If you did transcode (i.e. re-encoding a lossy to a lossy), then it is known that there is a generation loss which can make artifacts audible at bitrates which would be transparent if you did encode from a lossless source. If you are looking for advice on what to do about your lossy files, the general answer is “keep them”.
- If you are looking for general advice on settings to choose to minimize the filesize for given level of annoying artifacts, then I'd say the broad consensus is that VBR should be your first shot. Not equally universally agreed upon, but still likely to be best advice: if you worry that e.g. LAME V0 isn't good enough, then just stay lossless.
- If you have a lossless archive, but need space-saving settings for portable use, then why not start low? If it sounds annoying, just overwrite.
I don't know about battery life... depends on the device, I'm sure.
As for lossy encoder settings for best quality, it's a common question, and one that doesn't have a general answer that's going to be right for everybody. You have to generate the answer yourself, by testing. If you conduct proper ABX tests, you'll probably find that your suspicions about the placebo effect are quite correct, and that generally you don't need even half the bitrate or specific format you think you do, at least for getting "best quality". Oh, and you have to define "quality". What is that, for you?
The idea behind lossy is that the audio data can be simplified in space-saving ways that will quantitatively change the output, but that won't affect our perception at all, or that will affect it as minimally and least-unpleasantly as possible. The point at which a given piece of lossy-encoded audio, played through your equipment and heard by your ears, is indistinguishable from the original is the point of "transparency". Any knob-twiddling you do to "increase the quality" beyond that point is not really increasing the quality at all, because it's already at maximum for you. And as you might expect, where it is for you is not where it is for everyone else, and can be affected by the choice of audio to encode, your hearing loss, background noise, encoder, settings, etc.
Determining the point of transparency requires conducting a series of tests where you compare the original to two different clips: one is an exact copy of the original, the other is your lossy test version. These must all be volume-matched. In the test, you're forced to say which is which, even if you can't tell the difference. Do this enough times, and you'll have a meaningful score... 50% wrong answers (same as flipping a coin) means you couldn't tell the difference, less than 5% wrong means we won't argue with you anymore. foobar2000 is really good for ABX testing if you're doing it on your PC, and I believe there are ABX apps for smartphones, but it's going to be difficult to do a proper test with, say, an iPod Mini.
Its hard for me to determine since I think my ears have a placebo effect.
The first link I provided tells you how to avoid this very real and negatively influential phenomenon. Please read it.
In the second link please read about rule #8, in the first post as well as in the later post that describes the rule in more detail.
At this point I think you can safely discard any of the conclusions you've presented about different compression settings and how they affect sound quality.
I re-read the first link you sent me and very interesting stuff! It lamemans terms if I am perceiving it correctly it is almost like the human ear has no way of telling us about compression. It almost seems like it needs to be done with an abx test. "Science," as Thomas Dolby said. Since you seem like an expert on this field, does this mean that the only way to determine this is to break down every single song with this this test. It sounds like this would take decades with those with a large music collection. Greynol are you allowed to give your opinion on a matter such as Bitrate. I have no intentions of starting arguements as I can see you dont like those. I have read many post and once it seems if someone makes a statement and has no way to back it up, you will Greynol them..lol...I would love your opinion even if it isn't scientific on what bit rate that you like to stay at. Once again I dont know if you are allowed to do this according to the rules of the forums, but this seems like one of the better forums on the web when dealing with mp3's. I have so many lame post on sites where people argue about 192 vs 320. I am not interested in that garbage. I am interested in what the experts use when they are using MP3's if you even use them at all. Do the experts or yourself find any difference with lossy vs lossless without the help of technology. If you are allowed to make a statement that is pure opinion and what you use in the real world I would love to hear it. I figure take the advice of those that know, not because some extemist says 320 or die
I don't know about battery life... depends on the device, I'm sure.
As for lossy encoder settings for best quality, it's a common question, and one that doesn't have a general answer that's going to be right for everybody. You have to generate the answer yourself, by testing. If you conduct proper ABX tests, you'll probably find that your suspicions about the placebo effect are quite correct, and that generally you don't need even half the bitrate or specific format you think you do, at least for getting "best quality". Oh, and you have to define "quality". What is that, for you?
The idea behind lossy is that the audio data can be simplified in space-saving ways that will quantitatively change the output, but that won't affect our perception at all, or that will affect it as minimally and least-unpleasantly as possible. The point at which a given piece of lossy-encoded audio, played through your equipment and heard by your ears, is indistinguishable from the original is the point of "transparency". Any knob-twiddling you do to "increase the quality" beyond that point is not really increasing the quality at all, because it's already at maximum for you. And as you might expect, where it is for you is not where it is for everyone else, and can be affected by the choice of audio to encode, your hearing loss, background noise, encoder, settings, etc.
Determining the point of transparency requires conducting a series of tests where you compare the original to two different clips: one is an exact copy of the original, the other is your lossy test version. These must all be volume-matched. In the test, you're forced to say which is which, even if you can't tell the difference. Do this enough times, and you'll have a meaningful score... 50% wrong answers (same as flipping a coin) means you couldn't tell the difference, less than 5% wrong means we won't argue with you anymore. foobar2000 is really good for ABX testing if you're doing it on your PC, and I believe there are ABX apps for smartphones, but it's going to be difficult to do a proper test with, say, an iPod Mini.
I am so sorry with all my rambling about quality and I never stated my meaning. Quality meaning staying as close to the original musical source (CD or WAVE File) as possible even if it is lossy.
ThomasG3rd:
- Your original posting contained a few subjective opinions of the type that this codec of that bitrate was better than (“outshine” is quite a strong word) some competition. As long as those statements refers to sound quality (rather than e.g. to battery life), then they are per the terms of service not welcome on this forum unless backed by evidence (see the TOS) or infeasible to test (like “I had to rebuild my house after the fire, but at least I got rid of that annoying boom at 67 Hz” would be a statement way beyond your ability to back up by a listening test). Of course moderators will likely not knock you for stating “I tried to save space by going mono, but that sounded too annoying”.
- If you did transcode (i.e. re-encoding a lossy to a lossy), then it is known that there is a generation loss which can make artifacts audible at bitrates which would be transparent if you did encode from a lossless source. If you are looking for advice on what to do about your lossy files, the general answer is “keep them”.
- If you are looking for general advice on settings to choose to minimize the filesize for given level of annoying artifacts, then I'd say the broad consensus is that VBR should be your first shot. Not equally universally agreed upon, but still likely to be best advice: if you worry that e.g. LAME V0 isn't good enough, then just stay lossless.
- If you have a lossless archive, but need space-saving settings for portable use, then why not start low? If it sounds annoying, just overwrite.
Yes it was an opinion of mine with the word outshined. That was then though..lol..That was my hardhead days when It was 320 or die..I have tried to be open to the experts at what is the best way to go to achieve the quality of the orginal CD to the best ability by going lossy (my mp3 players dont support wav, flaq) I figure what is the point of using high bit rates unless they are needed. Eats up space and I think it eats up battery life. Mine Mp3 player did with wma's but thats only on my player. I cant tell you it there is any real evidence of this expect word of mouth from others
The most common answer to the question, "which lossy settings should I choose in order not to waste space?" is the following:
Pick a few tracks that are representative of your collection and encode them at a relatively low bitrate ~100kbits (for mp3) and see if you can ABX them from the original lossless source using foobar2000's ABX comparator. Increase the bitrate until you can you can no longer distinguish a difference.
For my own tests, I was able to ABX Lame at -V5 and sometimes -V4, so I ended up choosing -V3 in order to give me some margin which I felt that I could afford. Someday I might try this with AAC since it seems likely that it will save me more space, but as of yet I haven't seen the need.
a) With your bitrate settings chance is very low that you are able to hear differences. That's why greynol asked for how you determined quality.
b) The right way to check whether or not there is an audible difference to the original is by doing an ABX (blind listening) test. Probably the easiest way to do this is to use the ABX tool which comes with foobar2000. foobar is also a great tool for doing audio conversion and things like that, or to just listen to the music.
c) Disabling bitreservoir is a bad idea. Bitreservoir always has a positive effect on quality.
d) There is a widespread misconception about bitrate. In an mp3 file there are two kinds of bitrates. The most important bitrate is that of the audio data. But the audio data are packaged into frames (containers for the audio data) which have their own bitrates (which corresponds to the size of these containers). Bitreservoir allows audio data to spread over several frames. That's why there is only a loose relation between audio data bitrate and frame bitrate. When you demand Lame to keep bitrate at 256 kbps or above this is a specification for the frame bitrate, not the audio data bitrate. When you additionally disable bitreservoir you don't change audio data at all. You just use bigger containers for the same audio data. You have a lot of unused bits in your files.
You can figure that out by using the mp3packer tool on your files. You'll get smaller files with exactly the same audio data (to ensure yourself that I'm not talking nonsense you can use foobar's bit-compare track tool).
If you keep bitreservoir enabled this can have a positive quality effect when keeping (frame) bitrate at 256 kbps or above because this way the amount of data available in the bitreservoir will be a bit higher than when using the default settings. But you should use mp3packer afterwards to squeeze the many unused bits out of the file. And you should use 320 kbps as the general frame bitrate - when using mp3packer afterwards average bitrate won't increase.
e) With very high bitrate settings as you use them audible deviation from the original is very rare but it can happen. The essential question is how to deal with this situation, and you will find two groups of people here which often can't understand the other group's attitude. The probably major group feels like 'I don't care about rare events when quality isn't great. Usually quality is perfect, and when it isn't it's usually not annoying.' These people usually use -V2 or similar according to their personal attitude and needs. Sometimes they are additionally proud of going this way because they use their bits very effeciently and think it's a misuse of a lossy encoder to make use of it in a less efficient way.
Compared to this group the other group is a bit paranoid. Audible issues even when they are rare give them an uncomfortable feeling. They want to be very much on the safe side. Usually they still care about file size, but to a very minor extent especially as these days sufficient storage space usually is available or can be made available even on portable devices at low price.
I personally belong to the paranoid group, and you sound very much like you do too.
My advice: use lame3100i, a functional extension (http://www.hydrogenaudio.org/forums/index.php?showtopic=99483). lame3100i has a minimum audio data bitrate feature for instance, something that you obviously want to use but can't get with standard Lame which just allows for defining the mnimum frame bitrate.
For those who belong in the first category, you can always re-encode* using more liberal settings once you encounter a problematic sample. Even if this ever happens (NB my use of italics) then you know you haven't wasted any bits by bloating your collection by an additional 40(?), 50(?), 60(?), 70(?)%.
(*) EDIT: There is no law that states you have to re-encode your entire collection just because you found a problematic sample. You can simply re-encode just the ones that are problematic.
ThomasG3rd, I can suggest for the metal part of your collection to encode with -V 0 (no other settings) and use latest LAME. This is based on my personal experience. For a long time I used the setting -V 4, since it was transparent for almost all my metal music. However, I have since found a couple of problem samples (with some tracks from Nightwish, Amon Amarth, Keep of Kalessin,...) where I could achieve transparency only by using -V 1 or sometimes even -V 0. That led me to use only -V 0 from then on, just to be on the sure side. I know I am mostly wasting space, but knowing that I am doing all I can to avoid getting some problematic metal song to encode with artifacts is more important to me than the space those mp3s use, since space became really cheap and is not the issue nowadays.
I hope this helps...
About the listening tests, one Nightwish track (http://www.hydrogenaudio.org/forums/index.php?showtopic=55340) has it's own topic on this forum and I can post listening test from the other tracks if needed.
2.~245 with advance settings of lowest bit rate of 256 and highest of 320.
Why? Then you are crippling the VBR mode by only allowing it to use frames of two sizes.
I have bit resoviour turned off.
Why? Then you are crippling the VBR mode by not allowing it to divert bits from simpler frames to more complex ones.
This setting is good as I get a higher avreage bitrate, usually 295-305.
Of course you get higher mean bitrates. You have removed the encoder’s ability to use smaller frames when they are appropriate.
Pretty Close to the 320 mark I am sure that its just the same.
Of course it’s not “just the same”.
Big question about this is that if a song hits a passage of lets say 160, does the encoder still go that low since I set lowest to 256.?
How is this not obvious? Of course it cannot.
Did you use a dartboard to choose these settings? The defaults are the defaults for a reason, and I advise against changing them in any case, especially when the user does not seem actually to know what they’re doing.
Even if you were using sensible methods to increase the mean bitrate, that might be totally unnecessary if lower bitrates would equally provide perceptual transparency. I have to recommend further reading on both (1) LAME?s actual settings, rather than whatever oversimplified descriptions MediaMonkey presents in their stead, and (2) double-blind testing. Other users have already provided a lot of feedback, but you might want to start simpler, for example at the page about LAME on our wiki (http://wiki.hydrogenaudio.org/index.php?title=LAME).
I am so sorry with all my rambling about quality and I never stated my meaning. Quality meaning staying as close to the original musical source (CD or WAVE File) as possible even if it is lossy.
Talking of degrees of closeness to the source is irrelevant for lossy encoding (at least in terms of listening, which is what matters here). A lossily encoded file is either perceptually transparent or not. Beyond that, any increase in bitrate is a waste. Sure, there will always be rare exceptions that might not encode transparently at your chosen setting, but that doesn?t justify encoding everything unnecessarily high.
Besides, it seems futile to me to try to rank which settings in LAME are ?best? in that regard, as CBR and VBR have some fundamental differences in behaviour, so 320 kbps CBR may not be better than a high VBR setting (with a lower mean bitrate), and so forth.
I personally belong to the paranoid group, and you sound very much like you do too.
My advice: use lame3100i, a functional extension (http://www.hydrogenaudio.org/forums/index.php?showtopic=99483). lame3100i has a minimum audio data bitrate feature for instance, something that you obviously want to use but can't get with standard Lame which just allows for defining the mnimum frame bitrate.
Whilst it?s good that you provide this edited version and everything, I advise against recommending that new users adopt your personal methods simply because they remind you of yourself. Some rudimentary testing might do away with the OP?s concerns about bitrate, if not your own, so by recommending your edited version in lieu of such testing, you might be obscuring the truth of the OP actually experiences different bitrates perceptually. In other words, minimal mean bitrates may not be necessary, and I personally doubt they are for the reasons that I implied above about LAME usually being a sufficiently good judge of what an input stream demands.
foobar2000 is really good for ABX testing if you're doing it on your PC, and I believe there are ABX apps for smartphones, but it's going to be difficult to do a proper test with, say, an iPod Mini.
Not that difficult: have someone else create a playlist with a fixed (unknown to you) sequence of A and B, load it on the player and write down your guesses while listening.
foobar2000 is really good for ABX testing if you're doing it on your PC, and I believe there are ABX apps for smartphones, but it's going to be difficult to do a proper test with, say, an iPod Mini.
Not that difficult: have someone else create a playlist with a fixed (unknown to you) sequence of A and B, load it on the player and write down your guesses while listening.
Or have your computer make a set of files (test1, test2, test3, ...) randomly corresponding to A and B, saving a text file with the correspondence which you don't read until grading your test.
I don't know if it would make a big difference in this test, but humans are not real good at generating random sequences unless aided by something like a coin flip or dice.
The point is that it's way less convenient; doubtful very many people go to the trouble.
I personally belong to the paranoid group, and you sound very much like you do too.
My advice: use lame3100i, a functional extension (http://www.hydrogenaudio.org/forums/index.php?showtopic=99483). lame3100i has a minimum audio data bitrate feature for instance, something that you obviously want to use but can't get with standard Lame which just allows for defining the mnimum frame bitrate.
Whilst it’s good that you provide this edited version and everything, I advise against recommending that new users adopt your personal methods simply because they remind you of yourself. Some rudimentary testing might do away with the OP’s concerns about bitrate, if not your own, so by recommending your edited version in lieu of such testing, you might be obscuring the truth of the OP actually experiences different bitrates perceptually. In other words, minimal mean bitrates may not be necessary, and I personally doubt they are for the reasons that I implied above about LAME usually being a sufficiently good judge of what an input stream demands.
You as a member of group 1 I described above don't like that, sure. But aren't you a bit on a mission? Why shall I not give an advice to a new member who belongs to group 2 when I belong myself to this group, especially when the advice has very much in focus what the OP wants to do? Yes, my advice here is about my own Lane variant. I have rarely done so before, but when this variant perfectly matches the OP's intention why shouldn't I point to it?
As for a listening experience the issue at sec. 3.0 of problem sample eig is a good example. While I can perfectly accept that you (and other members of group1) don't see a problem here because when using standard Lame -V0 the issue is small and issues like these are very rare, I would welcome if you could respect other people's attitude towards things like this, no matter whether they are new or old members.
It all ends up whether we belong to group1 or group2. These groups have different things in mind for what is essential to them, and I think we should simply respect each other. There is no 'right' or 'wrong' here, just different personal needs.
[...] but when this variant perfectly matches the OP's intention why shouldn't I point to it?
I think db1989's point is that you're encouraging the OP to make decisions based on his arbitrary idea of what a suitable bitrate is, when he still has yet to perform any ABX tests. I doubt he would chide you for recommending your LAME extension to someone who has demonstrated an ability to ABX V0 in the default LAME branch, but when the OP has yet to even show that he has ABXed LAME at any level, suggesting that he automatically choose a version that will produce higher bitrates makes little sense and doesn't mesh with TOS8. The OP should determine if even V2 is necessary for him, let alone V0, before he considers moving to your extension.
OK, this makes sense.
I would welcome if you could respect other people's attitude towards things like this, no matter whether they are new or old members.
It wasn’t my intention to disrespect anyone or anything. Aleron Ives (thanks!) pretty much hit the nail right on the head in terms of interpreting what I actually meant.
As I implied, I think it’s a good thing that you offer an alternative version targeted to improve specific phenomena. If the OP does sufficiently extensive testing to find the default branch lacking in those areas, then great: he has another option. My point was that people generally should use the default release and settings unless they have a good reason to do otherwise, and recommending a different approach to new users – before they’ve even tried using the official release with recommended settings, never mind testing those objectively – just confuses things further and possibly suggests that the official branch has significant deficiencies that apply in general use.
First it needs to be determined whether the OP really needs a higher mean bitrate than offered by default switches. Enforcing a minimal size on frames doesn’t seem to me to be a useful thing if we assume that LAME is a good judge of instantaneous complexity: clamping the size at 256 kbps/frame or whatever is more likely just to inflate the size of less complex sections. Testing using more conventional settings and evaluating their perceptual performance seems to be more of a priority than worrying about enforcing a minimal mean bitrate; presumably the idea of the latter is based on concerns about quality, which aren’t admissable until testing has shown they exist.
I thank all for the comments and help. I was in no way offended by any of the comments (such as myself crippling the bit-rate by using 256) nor additional stuff that can be added by to lame. Some of the things suggested would be to complex for me, but I have a more of a feel of where to place my settings for lame. The overall "average" of opinion (and yes opinion to me is fine, since that is what I wanted from everyone.) is that ~245 bit resouviour turned on, no minium settings on bit rate. This seems the easist method for a newbie since yes I have no idea what the heck I am doing when it comes to playing around with the settings. I will download foobar is everyone seems to like this program. When I made a comment on "file size" I guess I really shouldn't care since i have plenty of storage space on my mp3 players. But I figured what is the point of wasting space on a 320KCBR when 1.It might be neccessary and 2.Lower bit rate at VBR might actually make the song sound better. Also some say that VBR makes battery life more effiecent. Not sure if this is true or any data to back it up, so that is another reason. I am looking forward to using Foobar when I get back home during the week. THANKS ALL, YOU ALL ROCK
The overall "average" of opinion (and yes opinion to me is fine, since that is what I wanted from everyone.) is that ~245 bit resouviour turned on, no minium settings on bit rate.
I honestly don't care what bitrate you use, but I have a hard time figuring where you came up with this as an average.
My interpretation of the OP’s references to that and similar descriptions was that MediaMonkey uses the mean bitrate to describe some underlying actual setting of LAME. As I implied when mentioning that before, I don’t much care for that sort of nomenclature due to the confusion that it causes when some files come out with a sizeably higher bitrate, which is of course the point of VBR!
.... where you came up with this as an average.
Possibly the HA Wiki entry for LAME, which shows 245 kbit/s as the "Target" for V 0?
As the OP is talking about '~245 bit resouviour turned on' (????) there's certainly still some confusion on his side. But as it sounds like he will turn to foobar as the conversion tool things hopefully will be easier to him.
Something we need to explain is that LAME has a number of presets ("V0" to "V10" or so; can't remember for sure). These presets include all of the encoder settings that the developers have determined through extensive listening tests to offer the most transparent (can't distinguish from the original CD) sound for a given target bitrate.
So, it is usually recommended that a user stick to the "v" settings and vary the "v" level to trade off your personal priority of transparency to you on your music vs. file size.
Stop worrying about settings and just choose a V level that works for you.
Personally I use V4 as it works for me 99% of the time transparently and I don't care if I get an occasional audible artifact. I can always revert to my lossless FLAC if I want perfection.
.... where you came up with this as an average.
Possibly the HA Wiki entry for LAME, which shows 245 kbit/s as the "Target" for V 0?
Yes, Thats what I meant for average, of where to set media monkeys settings. Since getting involved with using testers to test where each song should go (I just dont have the time) its easier to just let lame choose it for me. No min or max setting and enable bit res. V0~245 is the VO settign Media Monkey uses. It seems from what I learned (from the information that I have been kindly recieving) is that without testing each song with software that can tell me where to put the average BR once it analyzes a song, it seems to let Lame choose it for me. This way it seems like I get the best of both worlds. Forget wasting time playing with lames setttings, since as one person put it, unless you know what you are doing, dont touch the settings to "tweak." I know that Lame is going to choose my average bit rate at a higher level. Will some of it be overkill, I bet you it will, but as many of you said unless that song is analyzed, there is no way of knowing where the target bit rate should be. I will feel more confortable knowing that higher bit rates will be choosen (mind you at ~VO with no max or min setting and BR on, I am gettting an average of 265 with Metal, and 240-35 will classic rock and jazz. Some songs even hit below 200, thats fine, I am trusting lame. My songs won't be perfect, but they will be better then they have been since I am willing to listen to lame and let it decide, plus I dont have to worry about 320 being a means to an end. I even see some of you like V2 or V3 setttings. If I had better computer knowledge, I could edit the lame software by adding in the command lines that some of you spoke of, because it seems like that is how most of you are able to determine the BR. I have bought songs that were VBR's before and I have some Christmas music that is at like 105. Sounds fine to me, but it never bothered me at that bit rate since 1.It was Christmas music, it will get played 2x a year, or 2.It was Perry Como or Bing Crosbey and I highly doubt that Silver Bells needed a whopping BR. I will download Foobar and use that when I get a chance, since it might be better then mediamonkey for ripping CD's. I wish there was software out there that you could just drag your CD songs on, and it will automatically tell you what setttings of lame to use, but it doesnt seem that exist.
As the OP is talking about '~245 bit resouviour turned on' (????) there's certainly still some confusion on his side. But as it sounds like he will turn to foobar as the conversion tool things hopefully will be easier to him.
Im sorry, I think I confused you by not being clear. When using media monkey, and you choose to use VBR, it gives you a number of setting. VO~245 is the highest setting of VBR. Then there is a little box that says advanced settings. If you check it it lets you choose max and min bitrate. It also has a checkbox for enable bit resouviour and another checkbox for enable iso. Since people said dont tweak unless you know what you are doing with lame, I just choose the VO and check the enable BR, with no max or min settings. I stated below that this still might be overkill, but without a good conversion analyzer or having to analyze every single song, it seems like its better choosing these settings, then me playing with lame like I was before this forum started. I was messing around with it so much, unhappy as a song would dip below 224 because of my OCD when it come to my settings. There still where be times that I choose the higher VBR then needed, but until I get more time and a much better understanding of lame, its a start, so am I confused about lame still. But I know I shouldnt be messing with settings unless they are presets, such as the ones I described to you.
As the OP is talking about '~245 bit resouviour turned on' (????) there's certainly still some confusion on his side. But as it sounds like he will turn to foobar as the conversion tool things hopefully will be easier to him.
Just tried Foobar..much easier..no BS settings to tweak around with, but had to download a version of lame..choose version3.99.5 Doesnt have ten million things to do, i just choose vo vbr and thats it, even though Foobar says 320CBR is highest quality, hmmmm, not from what I read from here. There are no options for max or min br, bit resouviour, stereo or joint stereo, and the songs are comiing out great
It would serve you well to empower yourself by conducting your own ABX tests. I bring this up because of the discussion you bumped on the bit reservoir. Instead of relying on information that may or may not be reliable, you could easily test the quality of your encodes for yourself.
I think mp3 isn't a good choice if your very OCD prone even though the high bitrates have been more tuned in recent releases and its expected to produce acceptable results even in bad cases. Other options are MPC Q6..7 , Vorbis Q7..8, & AAC. Consider even moving to lossless. Get a player (its not hard there are much more options like android devices) that plays flac / wavpack etc and stop transcoding .
Just tried Foobar..much easier..no BS settings to tweak around with, but had to download a version of lame..choose version3.99.5 Doesnt have ten million things to do, i just choose vo vbr and thats it, even though Foobar says 320CBR is highest quality, hmmmm, not from what I read from here. There are no options for max or min br, bit resouviour, stereo or joint stereo, and the songs are comiing out great
I still don't think you've understood what people have told you.
Why do you think you need to play with the default settings? How is foobar much easier? It only became complicated once you decided to start messing with the advanced options. Foobar's slider is a general slider over a range of settings, the lowest option doesn't even mention quality, just "Smallest file".
As you've been repeatedly asked to do, go do some ABXing, you might find they will "come out great" at a lower setting.
I think it's a bit unlucky that now that the OP has found his way which is very much what was suggested to him and now that he his happy with the results more suggestions come up which are able to rise doubts to his new way of doing things.
It's all valid opinion, but just that. Let's be happy that he found his way which I think all of us beleive to be a good one.
@ThomasG3rd:
Sounds like you still are a bit worried about bitrate. You shouldn't. Think of a track which contains large parts of very low volume (silence for instance). Few bits are needed for a perfect quality here. Average bitrate will go down very much when using VBR, for exactly this very good reason.
Also don't worry too much about having read 'CBR320 is best'. It's more of a traditional saying which was true in earlier years of mp3. Today it's a very doubtful statement, and has already been proven wrong at least for some tracks.
You chose the best quality VBR level -V0, and you'll get great quality.
more suggestions come up which are able to rise doubts to his new way of doing things.
Would you mind being more specific as to what these suggestions are?
My impression is that most of the confusion comes from the fact that the OP has/had the impression that a high bitrate is desirable, when really the opposite is true.
more suggestions come up which are able to rise doubts to his new way of doing things.
Would you mind being more specific as to what these suggestions are?
shadowking's remarks in the first place. But I didn't want to be so specific. I just wanted to say that it would be good for a new member who has found his way (and still has some trouble finding peace of mind with it) not to be further confused.
With all due respect, based on your last two posts, it seems that you're the one who is on a mission.
If you can truly detect lossy artifacts then it's a good thing to consider trying more modern codecs which are not hindered with the same limitations as mp3; limitations which can't always be solved by bloating the bitrate.
The suggestion to use lossless also holds merit regarding the circumstances (operating on fear; refusing to perform personal listening tests).
Thomas, the problem that you are having is that you are confusing bitrate with quality. They are not the same thing! Some passages of music are difficult to encode, and some are easy. When the bitrate drops temporarily within a song, it is not a drop in quality. It simply happens that it took less space to meet the quality threshold (e.g. V2) that you specified. As an obvious example, digital silence can be encoded at extremely low bitrates (32kbps) without any loss of quality, subjective or objective!
The fact that a "song would dip below 224" (your words) is a byproduct of that passage of the song being simpler in terms of harmonic content. It is the exact same audio quality as the rest of the song.
I have done some listening test and yes I have found some songs that dont need a high bitrate. I figured out why the Animals "House of the Rising sun" wouldnt go into VBR mode as I had the wrong settings orginally when using media monkey. The average BR for House was 165 when using foobar and 144 when using mediamonkey both at V0. No audiable difference from my orginal 320CBR version. Funny, that song must not use a BR higher then the 160's. I played it in Foobar and the results were 160's to the 130's for the BR. As for bit resouviour, I can't tell any difference. There is no option to enable or disable it in Foobar and when I ripped Steely Dan's "Black Cow" @V0 came out at 265 and in mediamonkey with bit resouviour enabled it came at 245. I cannot tell a difference between the two. Both songs sound great, the highs are nice and smooth with no harshness and the bass parts of the song hit acuruately and deeply. So the listening test are going good, as if I am using the best settings, well the from my ear standpoint, yes, as for a scientific standpoint I have no idea since I dont have software to determine this. But I am glad that I got responses to this post. My whole point was trying to get sound that was as true to the souce (cd or lossless audio file) as possible in mp3 form, without having to use riducuolous bit rates. I had no idea what VBR was about and I have gained more knowledge. I know that I might be over or under the BR, but I feel much better that I dont have to spend hours tweaking setting just because a song feel below 220br. Thats the habit I was trying to break and even though almost everything here posted are peoples views or opinions, that is fine, since I know that all of you have been using the VBR a lot longer then I have. I am sure some of you are audiophiles or picky about your music. Some of you might care less, but the point was an overview of what people were using. I never mentined anything about the equipment that I was listening my music through since this can be dangerous topic. I dont think the audio quality changes just because of your reciever or computer speakers, well at least not scientifically. Some have mentioned that unless you use high end equipement its pointless to use high bit rates (and I have seen many including Graynol rip them and prove it with evidence that equipment changes the file) One pair of speakers might be good for some and others could think it sounds like a tin can. I will post more as I learn more, and any other ideas on the subject I would love to hear.
So you're using the ABX utility within foobar2000 to determine if you can tell the difference between two versions of the same song? If so then that's great. My only other comment is that you should probably test the lossy encode against the lossless source used to create that encode. If you can demonstrate that you can differentiate between two lossy encodes it doesn't tell you which one, if either (or possibly both, though I doubt it), is transparent to the lossless source.
Just tried Foobar..much easier..no BS settings to tweak around with, but had to download a version of lame..choose version3.99.5 Doesnt have ten million things to do, i just choose vo vbr and thats it, even though Foobar says 320CBR is highest quality, hmmmm, not from what I read from here. There are no options for max or min br, bit resouviour, stereo or joint stereo, and the songs are comiing out great
I still don't think you've understood what people have told you.
Why do you think you need to play with the default settings? How is foobar much easier? It only became complicated once you decided to start messing with the advanced options. Foobar's slider is a general slider over a range of settings, the lowest option doesn't even mention quality, just "Smallest file".
As you've been repeatedly asked to do, go do some ABXing, you might find they will "come out great" at a lower setting.
Because Foobar doesnt have any advanced settings. It was complicated because in the beginning I mentioned newbie at VBR. When you never used a format and there are options such as advanced settings (such as MEdiamonkey, Goldwave), they can be confusing. As I stated, I understand what people are telling me which is a number of things:
1.Bit rate at high levels dont always mean quaility.
2.Dont play with advanced settings let lame choose the quality for you
3.Bit resoviour seems to be the overwhelmiing choice, BUT I cannot tell the difference by listening test, but I leave it enabled if Im using mediamonkey.
4.VBR seems to be favored over CBR even if the CBR is higher, VBR such have a better quailty as close to the orginal source at a lower bit rate.
5.One thing that I stilll dont understand is that some say BR doesnt mean quality (at any level.)There had to be a reason that there are "choices for this" Otherwise wouldnt it just be easier to have a standard default BR in lame lets say at 64 and it would all sound the same, thats where my confusion lies, so if you can help me in this area please post.
Has anyone mentioned the FAQ?
Has anyone read the FAQ (or the wiki it leads to on this subject)?
It's quite clear...
http://wiki.hydrogenaudio.org/index.php?ti...ncoder_settings (http://wiki.hydrogenaudio.org/index.php?title=LAME#Recommended_encoder_settings)
btw, your "Animals" track is probably mono, which dramatically reduces the bitrate required, hence the VBR modes use a lower bitrate.
Of course, on a given track, a higher bitrate means better quality - but that quality may be beyond what you can actually hear, and different tracks needs different bitrates to reach the same perceived quality - hence VBR.
Cheers,
David.
even if the CBR is higher, VBR such have a better quailty as close to the orginal source at a lower bit rate.
I would not go so far as to imply that it impossible for CBR to give better quality than VBR.
By quality, I mean audible sound quality. Better quality past the point of transparency is meaningless from the standpoint of audibility (which is the name of the game, is it not?).
5.One thing that I stilll dont understand is that some say BR doesnt mean quality (at any level.)There had to be a reason that there are "choices for this" Otherwise wouldnt it just be easier to have a standard default BR in lame lets say at 64 and it would all sound the same, thats where my confusion lies, so if you can help me in this area please post.
A main misconception here is to consider bitrate a target, when aiming for transparency. What a lossy perceptually codec in a VBR mode target for is quality, not bitrate, the latter being a "side effect" of the first AND of the complexity of the source signal AND of implementation optimizations.
Quality is "qualitatively" defined as how much the encoder is allowed to degrade the original signal with respect to the perceptual model implemented. In other words, a quality switch fixes the amount of information the perceptual model allows to throw away to reach a desired level of "closeness" to the source. If, again, a listener aims to subjective transparency, he has to find his own quality level sweet spot for the selected codec and let the encoder do his job, forgetting about all this bitrate stuff.
5.One thing that I stilll dont understand is that some say BR doesnt mean quality (at any level.)There had to be a reason that there are "choices for this" Otherwise wouldnt it just be easier to have a standard default BR in lame lets say at 64 and it would all sound the same, thats where my confusion lies, so if you can help me in this area please post.
There are three basic factors that determine quality in MP3 encoding
1) The scientific sophistication of the encoder. LAME is very mature and has been worked on for a long time. It's fair to say that it is "as good as it gets" for MP3, barring some scientific breakthrough. Older encoders can be worse. A notable example is the "Blade" encoder. It was really, really bad. On the same song, 256kbps CBR Blade will sound
worse than 128kbps CBR LAME.
2) How difficult the song is to encode, and this varies not only among songs but also varies within the song itself. Crash cymbals are a lot harder than violins, for example. It is difficult to quantify complexity in terms that make sense to a person.
3) The bitrate. Higher bitrates allow for more quality, but the bitrate itself is only a partial indication of the audio quality. In the case of VBR, you need to be careful about comparing quality based solely on bitrate because the bitrate fluctuates throughout the song. In VBR mode, encoders try to maintain a constant audio quality, and the bitrate can fluctuate. In CBR mode, the encoder maintains a constant bitrate and the quality can fluctuate. CBR only makes sense for fixed-bandwidth applications like streaming internet radio, satellite radio and cable TV. VBR mode is sometimes called "constant quality" to better explain the difference between it and "constant bitrate".
As for bit resouviour, I can't tell any difference. There is no option to enable or disable it in Foobar and when I ripped Steely Dan's "Black Cow" @V0 came out at 265 and in mediamonkey with bit resouviour enabled it came at 245.
The bit res
ervoir is enabled by default and should not be disabled without a valid technical reason. Myself and at least one other member have already explained why the feature exists and why disabling it is inadvisable, so I’m not sure why you’re still concerned about it. foobar2000 will also leave it enabled; the different bitrates that you report probably arise from the two programs using different versions of LAME, not any change in the behaviour of the bit reservoir.
I cannot tell a difference between the two. Both songs sound great, the highs are nice and smooth with no harshness and the bass parts of the song hit acuruately and deeply. So the listening test are going good, as if I am using the best settings, well the from my ear standpoint, yes, as for a scientific standpoint I have no idea since I dont have software to determine this.
Do you mean your listening tests weren’t double-blind? Your less than scientific descriptions of the tracks suggest that they might not have been. To be fair, we’re not really going to quibble very much if your tests were sighted and you aren’t reporting a difference, other than mentioning that they’re not very reliable. When people start claiming differences from sighted tests, often with similar language, it’s a different story.
But I am glad that I got responses to this post. My whole point was trying to get sound that was as true to the souce (cd or lossless audio file) as possible in mp3 form, without having to use riducuolous bit rates. I had no idea what VBR was about and I have gained more knowledge.
Good to hear.
Some have mentioned that unless you use high end equipement its pointless to use high bit rates (and I have seen many including Graynol rip them and prove it with evidence that equipment changes the file)
I’m not sure what exactly the statement in parentheses is referring to, but since you mention it, I recall a recent allegation that the likelihood of identifying audible degradation increases with more advanced equipment… and I recall greynol responding that some users have observed exactly the opposite and that such such things should not be claimed as general trends when they are far from universally applicable.
5.One thing that I stilll dont understand is that some say BR doesnt mean quality (at any level.)
I think best is a demonstration:
I encoded Green Day's 'Wake Me Up When September Ends' using Lame 3.99.5 -V0 and
a) keeping bitreservoir on (which should be done!) -> Wake_br_on.mp3 (266 kbps)
b) switching bitreservoir off (nonsense of course) -> Wake_br_off.mp3 (284 kbps). Don't expect that the higher bitrate means a higher quality! The higher bitrate is solely due to unused bits.
c) I squeezed the unused bits out of Wake_br_off.mp3 using Omion's great fast and lossless mp3packer tool -> wake_br_off_mp3packer.mp3 (260 kbps !!!). So even with the wrong focus on bitrate you see that Wake_br_on.mp3 is best!
Probably you doubt that the audio data of Wake_br_off.mp3 and wake_br_off_mp3packer.mp3 are identical. But this can easily be verified using foobar:
(http://dl.getdropbox.com/u/2681777/test/br.gif)
Thats really cool, I hardly got a chance to mess around with foobar, but I would love to test some of my tracks, of course when Im done going through the painful labor of re-ripping my cds and flaqs (well converting my flaqs.)
Since I am cannot re-rip my whole collection, (some of my files were ripped from cd's i dont own anymore) anyone have suggestions if it would be OK to change a wma over to a mp3. My wma collections ranges in 256 to 320cbr mode, non-protected. Since this is a VBR topic, would it be OK to convert the 320's over to MP3 in VBR mode, or is the going to screw up the final result. (and use V1 or V2 for the 256's) The only problem with this would be the ones at 256. I would have to set the encoder to a 256bit rate max, because 320 doesnt exist on a 256. I know im cutting down on the orginal quaility, (well i have to assume i am considering going lossy to lossy) and Im not sure if the final result will result in a messy mp3 or I could go with a ABR? NOt sure so ANy input would be great
Since I am cannot re-rip my whole collection, (some of my files were ripped from cd's i dont own anymore) anyone have suggestions if it would be OK to change a wma over to a mp3. My wma collections ranges in 256 to 320cbr mode, non-protected.
Converting between lossy formats reduces quality. However, since the source material is fairly high bitrate, you may not be able to notice any difference. Try it and see.
The only problem with this would be the ones at 256. I would have to set the encoder to a 256bit rate max, because 320 doesnt exist on a 256. I know im cutting down on the orginal quaility, (well i have to assume i am considering going lossy to lossy) and Im not sure if the final result will result in a messy mp3 or I could go with a ABR? NOt sure so ANy input would be great
This does not really make sense. Are you thinking that for some reason you should try to match the bitrates between the source and destination files? If so, there is no reason to do that. The second encoder has no idea about the first. It just tries to encode the file you give it (including any loss of quality from the first encoder) with minimal audible difference. If you want very little audible difference, use high quality VBR. If you think that the file is already degraded, perhaps you don't need very high accuracy and can use a lower setting. Try it and see.
I don't believe it's controversial to say that you stand to lose more audible quality when transcoding transparent mp3 to a second generation mp3 than converting lossless to mp3 using the same settings due to increased temporal smearing.