which bitrate?

[The Man With Brass Eyes]

ripping CDs?  which bitrate.
I'm guessing 128 is a pretty good minimum for most pop/rock music.

however, would anyone advise that I should up the encoding to say, 192kbps?
(Incase I ever listen to these songs on a portable MP3 player)

it's a bland thread I know... I can always come back and make it more uninteresting.  it's a challenge, but I'll try.

[kidsick5000]

192 is standard 'best' quality'.
depends if you plan to share or not.

192++ sould get see you fine

[The Man With Brass Eyes]

does the type of song (ie classical) benefit from a higher bitrate?
I get the impression classical has "more of a range" than yer pop rock racket, which will be good enough at 128.

[Rats]

Definately 192 or above, 128 shits all over the drums

[Jaffa The Cake]

What Rats says. I wouldn't encode an MP3 at less than 192kbit, and I normally go for VBR.

Here's my official 128kbit vs 192kbit comparison:

Bass drum:

192kbit: "Boom"
128kbit: "Flump"

Cymbal:

192kbit: "Tisssss"
128kbit: "Trischholoch"

The abilities of a compressor differ from song to song, more than genre to genre. Musically complex / intelligent tracks don't necessarily require a higher bitrate. For instance, the MPC codec (similar, but superior to MP3) can reach transparency (sound like the original CD track) at ~110kbit with some portishead tracks, yet a shitty smashy punk track would require ~180kbit.

[Borboski]

Yes PLLLLLLLLLEEASE rip at 192 it makes me so sad when I get mp3s at 128, I know I'm probably tricking myself but it always sounds much much worse.

Also, can anyone tell me why I often find files on Soulseek at bitrates of 400-500 kpbs? I killed loads of CD's trying to put albums on at this size. Why does it happen?

[9]

I use 192 for ripping. Jaffa and Rats explained why.

I've started compressing down to 64 when I transfer files to my mp3 player, just so I can squeeze more on. It probably removes a lot of the definition, but to be honest I don't notice when i'm walking down the high street.

[9]

and I normally go for VBR.

What's VBR?

[Jaffa The Cake]

For those naughty people who download music and want high quality, download albums off Emule. They're zipped / rar / ace-ed together in one archive, and normally ripped & encoded well.

[mayer]

and I normally go for VBR.

What's VBR?

Variable Bit Rate... the bit rate changes for every second of the song depending on how loud it is etc. quieter bits need only a lower bitrate to be of a certain quality.

[Jaffa The Cake]

What's VBR?

Rather than use 192kbits a second each second, it uses what it needs depending on the quality it's trying to achieve. So, it won't waste 192kbit on silence, it'll use near zero. It won't use 192kbit on a second if it'll sound shit, it'll go higher.

Edit:

Yeah, um. As mayer said. However, it's not just a quiet / loud thing as it can sometimes seem. It's the complexity of the sound. However, fast / loud songs tend to be complex (in terms of wave), like shitty smashy punk songs, while a smooth strings piece will require a lower bitrate.

[jutl]

Of course, 128 can be a great bitrate if you're using that marvellous better-than-mp3 open standard ogg. I tend to splurge and do oggs@192kbps, which sound lovely and don't eat up too much of your SD Card.

[Jaffa The Cake]

Ogg is useless at 128kbit, just less useless than MP3.

This is why I use MPC, it's aim was different to all other codecs in development at the time.

Most codecs - "Let's make 128kbit sound less bad than all the other codecs"
MPC - "Let's make something that sounds transparent at ~170kbit"

[The Man With Brass Eyes]

so which encoders (good that is) are there for ogg?

which commercial encoders feature vorbis ogg?

[gazzyk1ns]

I wrote this a while ago for the "How to" thread:

http://www.cookdandbombd.co.uk/forums/index.php?topic=1355#31759

To possibly save people time whilst reading it, the bit in the big quote box is only about encoding to MP3 with LAME, so skip that bit if that's not what you were after.

A good ogg encoder can be found on the relevant Rarewares page here. That page looks very confusing but all you need is one from the top box there, either the generic one or a compile which corresponds with your CPU type if you prefer.

Is that what you use Jutl? Maybe there's a better one, I just remember finding that quite good when I tried it a while ago.

EDIT: To be clear, although in my big linked post in the "How to" thread I say ogg is a format invented for lower bitrates, I certainly wouldn't contradict anything Jutl said up there.

[Mr Colossal]

Yes, 192 is the first  bitrate to reach  what is generally classed as 'CD quality. '

Apparently the human ear can't tell the difference between 160 or anything higher than 192, but you may want to go one higher if your recording from vinyl, as it has a slightly wider frequency range. Again, whether this is noticable to the human ear or not is argueable.

 I have heard some better sounding 128 encodes, though i tend to notice a light phase/tinny sound to the higher frequencies.

Also i've noticed that tracks encoded at 128 distort at a much lower volume than their higher encoded counterparts.

[gazzyk1ns]

Yes, 192 is the first  bitrate to reach  what is generally classed as 'CD quality. '

Apparently the human ear can't tell the difference between 160 or anything higher than 192, but you may want to go one higher if your recording from vinyl, as it has a slightly wider frequency range. Again, whether this is noticable to the human ear or not is argueable.

Some people are proven to be able to tell the difference between MP3 at 192 and above. They are in a minority though. I can tell the difference between a LAME mp3 at 160 and the original wav, but not 192.

To see what you can and can't identify as sounding different from the original then you need to do a blind ABX test, a full description/explanation is here. Some of the wording in there is a bit... unusual but don't think that it's written by an idiot or was done in a hurry, pio there is French. There's a lot of concentration on the probable outcomes and how results can/should be interpreted there, that's because there's a lot of debate and misunderstanding over on Hydrogenaudio about how to interpret the results of some of their mass tests carried out in the past. Therefore you might want to scan-read paragraphs concerning those things and just get on with grabbing and using testing software, linked to right at the bottom.

[Jaffa The Cake]

Apparently the human ear can't tell the difference between 160 or anything higher than 192

Sounds like a made-up fact to be honest. It depends on the encoder. I've known 128kbit MP3 encoded with LAME to sound better than 192kbit MP3s encoded with BladeEnc.

Also i've noticed that tracks encoded at 128 distort at a much lower volume than their higher encoded counterparts.

Depends what you mean by distort... the distortion's most likely created by your amp & speakers rather than the MP3. But if by distortions you mean "artifacts", then (as with all lossy audio compression) they're more noticable at higher volumes.

[Mr Colossal]

That too, i have a few Mix sets downloaded at 128kb and when i play them loudly it sounds terrible, compared to Mp3s recroded at 192 which sound just as good blaring...

Though this might be down to me altering the E.Q of the file, sometimes I've thought a particular mix sounded a little flat through my standard computer monitors, and aplified the lower end... Of course when i play it on my hi-fi, complete with subwoofers, sometimes i've overshot it turning them into wind machines.

I used to be one of those 'bass freak' type people when i was about 16... I've seen the error of my ways now.

[Jaffa The Cake]

Ahh. I'm one of these purist freaks who doesn't believe in EQs and wants to rid the world of things like "Dynamic Compression".

Having said that, the 5.1 to stereo downmixer is good in PowerDVD... truesoundXT I think it's called. Very good at getting a subwoofer effect from a pair of decent speakers.

[The Man With Brass Eyes]

I have to admit - listening with headphones on a computer - I can't especially tell the difference between 128 and 192.  I don't think I've done a direct comparision between the same song backtoback.
I'm listening to mainly rock/punk stuff these days.
Ok, I'm not listening to these files at mega-loud levels - so there's no distortion as of yet.

I'm wondering whether I need to rip my entire CD collection to the more desirable 192.

[Rats]

If it doesn't bother you then just do it from now on and keep what you already have. Mind you, every time you listen to 128 mp3 you'll be thinking about it now.

[The Man With Brass Eyes]

Mind you, every time you listen to 128 mp3 you'll be thinking about it now.

I am listening to the London Calling Vanilla Tapes so the quality does sound ropey.

[Marty McFly]

I have to admit - listening with headphones on a computer - I can't especially tell the difference between 128 and 192.  I don't think I've done a direct comparision between the same song backtoback.
I'm listening to mainly rock/punk stuff these days.
Ok, I'm not listening to these files at mega-loud levels - so there's no distortion as of yet.

I'm wondering whether I need to rip my entire CD collection to the more desirable 192.

there's no reason not to use 192kbps. storage space is as cheap as hell.

you may not be able to hear it NOW, but if you listen to more and more MP3s of varying levels, you'll soon be able to tell the difference between 128kbps and 192 or above.

here's a good guide to perfect MP3 ripping.

http://home.iprimus.com.au/alexanderino/cdex/

[The Man With Brass Eyes]

there's no reason not to use 192kbps. storage space is as cheap as hell.

I don't think it's so much to with space  - it's just the slower ripping speeds....

I've never really got on with CDEX - I tend to use Musicmatch.  what  other alternatives are there?

you may not be able to hear it NOW, but if you listen to more and more MP3s of varying levels, you'll soon be able to tell the difference.

I do listen to a hell of a lot of mp3s, but they are all at 128 - I've not really done a serious comparison with the higher (192) rates.  Maybe I'll do the pepsi kbps challenge later today.
...expect a post from me with the words "wow!! how wrong was I ?!!!!" later....

[The Man With Brass Eyes]

Right, I've just ripped the Von Bondies "Lack of Communication" album at 192 kb/s to do a direct comparison between the 128 version.

Tell you the truth, I'm finding it very very difficult to tell the two apart.
Maybe this isn't the best type of music to use for comparisions (the distortion & hiss are naturally part of the production style).

(or should I really be comparing the encoded file with the original uncompressed CD?)

can someone give me an indication of what (sounds/instruments) I'm looking out for.... or give me an album which is more useful to test my ears on.

maybe I need to visit the optician.

[Marty McFly]

i don't know if you can tell like THAT - in a pinch.

three or four years ago, i burned some audio CDRs of 128kbps mp3s that i had floating around, purely so i could listen to them on my CD player. at the time, i couldn't tell the difference between those and CD quality audio.

i listened to one of the CDRs a couple of weeks ago and straight away i could tell it was from mp3. especially the sounds of the cymbals. those are the main giveaway of a bad encoding - you get a swishy swashy sound that's horrible.

i think it's just something that your hearing builds up over time.. i dunno.

plus, as someone in this thread pointed out, what encoder you use has a great effect on the sound quality. i don't know what encoder ships with musicmatch.

here's a rather interesting test.. note that the 128kbps MP3s used in this test were made using VBR with LAME - which is, without a doubt, the best codec on the market.

some good selections of music in there to test, too.

http://www.rjamorim.com/test/multiformat128/results.html

[gazzyk1ns]

Yeah you should be listening to short samples and comparing it to an uncompressed wav of the same sample (well, to start off with, at least). Start with 128, verify that you can tell the difference (I'd be very surprised if you couldn't)... then see if you can tell the difference with a 160.

(On the samples I tested) I can tell the difference between a 128 and the original without headphones but you should be using headphones all the time for these tests, that way, when you fail to distinguish between the original and the compressed file, you know you'll be safe with that kind of bitrate no matter "how" you're listening.

The MP3 format has big problems with "fast attacks", for example, castanets. This page has a sample of some castanets and also explains what to listen for when trying to spot the different types of compression artifacts.

Use a programme like ABC/HR to do the tests so that you don't know if you're listening to the original or the compressed file... if you do know then there's a good chance of a placebo effect. The proggy looks really complicated at first but it's really not, you can use as many of those individual "consoles" as you want. That linked page has a comprehensive user guide but to simplify, you have a blue play butting marked "ref." - this is always the original, i.e. the uncompressed wav. You need this reference because the whole idea is to see if you can hear that the compression has changed this original. The other two play buttons, the black two which are side by side, are another copy of the original, and a copy of the compressed sample. You need to listen carefully to each of the black arrows, comparing them as and when you like to the blue one. If you can identify one of the black arrows as sounding different to the other black arrow and the blue, then you have identified that you can hear compression artifacts. Either that, or you are imagining things or your ears are getting tired after too much "concentrated listening"... you can tell if that's the case by your results of course. Knowing what bitrates you're looking at testing, you should have groups for 128, 160, and 192... i.e. use three of the eight test consoles.  You should, of course, use the same sample with all the bitrates.

After you've identified what sort of bitrate you begin to be unable to hear artifacts at, you might want to go on to do an ABX test (also explained on that same page) with just one bitrate vs. the uncompressed file. This is to determine how obvious the artifact(s) are to you at that bitrate... for example, after extended concentrated listening, you might have found that you still only JUST managed to hear a small imperfection at 160... so it's therefore fair to conclude that even so, you'll probably never hear the difference under normal listening conditions and so to go up to 192 would be overkill. To find out whether that's true or not, try to "ABX" the file a minimum of about 16 times and see how often you are successful. When thinking about your results, bear in mind that unless they're perfect one way or the other, you could well have been guessing... that's what all that "probability" waffle was about in the "What is a blind test?" FAQ I linked to earlier. Just remember that when doing the test, it's useless thinking "Well... I can't really tell the difference... so I'll just press this butting to complete the test...". If you can't tell the difference, then there's your answer.

All of that sounds like lengthy work, and it is - but it's the only way you'll be able to tell whether you can hear compression artifacts at given bitrates or not. If you  haven't got the time to do it then speculating and doing some tests which are going to be extremely subjective (and possibly impossible, like listening to whole tracks at a time) isn't going to help, you just need to do what Jutl describes he does - pick a bitrate you know you'll be satisfied with through experience of listening to it. If you're not really sure, then I'd say that if you're encoding with LAME MP3 or OGG, I'd bet all my possessions that 192 will be perfect for you.

Did you read that little bit I wrote in the "How to" thread on encoding MP3s? To use a constant bitrate is specifying an inferior quality, use VBR, specifically a preset. Use LAME 3.90.3 with the command line "--alt-preset 192". If your portable doesn't like VBR then fair enough, obviously, but most people just have an irrational fear of it. Strangely the same people often admit that ogg is excellent, and that's always VBR.

Some samples which have been picked out as being potentially tricky to encode can be found here, there's some Smashing Pumpkins and some Kraftwerk in there. Something like typical Aphex Twin will also be tricky to encode.

[The Man With Brass Eyes]

Phew a lot to take in there Gazz – thanks.  I'll give those tests a go.

Ah compression artefacts at given bitrates – I guess that's the deciding factor - whether you can spot the discrepancies... since it's lossy encoding  (...see I'm learning).

I'm still trying to determine which encoder Musicmatch 9 uses, and at which bitrate "style" I'm using (ie VBR or otherwise).  According to here, Musicmatch uses the Fraunhofer (FastEnc) and MP3Pro encoders.

I get the impression that VBR allows for greater accuracy (as it varies the quality based on the complexity of audio it's hearing), but with potentially larger file sizes than CBR.

Musicmatch appears to give a % slider for "customer quality (VBR)" with no option to choose kbps.  Do kb/s not apply for VBR encoding?   I suppose not if it's variable and isn't nailed down to a set rate.

[The Man With Brass Eyes]

One other question which has just bubbled up at me is

Should you (or when should you) need to go higher than a bitrate of 192?
If you're ripping uncompressed audio (CDs) - will 192 do for all intents and purposes?

(insert "I've got your number" joke here)