Article: In What Format Should I Rip My Music?

[haiku1]

The bit about VBR makes sense now. Thank you! :-)<br />

But I still don`t get your explanation about lossless.<br />

You say VBR is a lossy technique, but then you say ALL lossless formats are CBR.<br />

?????<br />

<br />

By the way, I thinks it`s very interesting that Apple claims VBR to be of higher audio quality then CBR, when the iTunes+ files are all CBR, according to the guys on Hydrogen Audio.<br />

<br />

What`s your opinion, Idolse?

[PeterSt]

... is that that definition on Hydrogen is not correct. Or I don't understand how to interpret it.<br />

<br />

The phenomenon "Variable Bit Rate" inherently belongs to MP3-like formats.<br />

With a given "quality" (indeed put upside down) the number of bits used vary. However, the quality never is losless in this case. It just isn't because it doesn't work like that. Not with "VBR".<br />

<br />

Compare FLAC. FLAC is "variable bit rate" just the same, but is not called like that because "VBR" denotes lossy. It's just in the book (but don't look at the Hydrogen book). A classical FLAC may compress to 60-70% while rock-like compress to 40-45% ussually. This implies a variable bitrate, which again, is not called like that.<br />

<br />

A Constant Bit Rate is not per definition losless. Look at "CBR" MP3. Constant bit rate, but sure lossy.<br />

A WAV file though, is constant bit rate also, but is losless. And it would not be called "CBR" because again it denotes "MP3".<br />

<br />

It is all very easy to understand, as long as we're not reading the wrong things on the Internet ...<br />

<br />

Peter

[haiku1]

but in which book can I get all the information? So far, I`ve got only your opinion, but thanx anyway! :-)<br />

I gather it`s not that easy to understand if you look at the posts from Idolse, yourself and me.<br />

Maybe we should start a thread about it?<br />

What makes me wonder, doesn`t Chris have any opinion about it? Chris, please.......!<br />

[PeterSt]

You're right. That book probably doesn't exist.<br />

If I had to again summarize it, I would say :<br />

<br />

The variable bit rate in any lossless format says/does nothing. It has no function other than compressing.<br />

<br />

A variable bitrate in a lossy format, indicates the method of compressing, and it has a supposedly better result for (sound) quality for the achieved compression factor.<br />

<br />

and<br />

<br />

The constant bit rate in a native lossless file (like WAV/AIFF) is just derived from a constant sample rate at recording.<br />

<br />

A constant bit rate in a lossless compressed format does not exist.<br />

<br />

A constant bit rate in a lossy format emerges from cutting out the higher frequencies, therewith requiering a lower virtual sample rate.<br />

<br />

also<br />

<br />

Supposing a 16/44.1 original file, no matter in what format that file appears, it is without upsampling etc.) played back at 16/44.1 again, showing a 1411000 bitrate. It can't be done otherwise.<br />

<br />

Supposing a normal DVD format this counts the same. BUT, because DVD inherently anticipates on VBR (this is just always the case) a DVD can easily be recorded in a higher compressed bitrate than normal. These are "superbit" DVDs, and they play in any DVD player.<br />

Unlike the 16/44.1 example, they play back at a higher bitrate than normal (but "normal" does not exist for MPEG2 (et al) movies. This is because CBR does not exist for movies.<br />

<br />

Hope it helps a bit. :-)<br />

Peter<br />

<br />

[haiku1]

it would, Peter.<br />

Perhaps you could explain just a bit more, please. I`m really interested!<br />

I understand that you say a lossless format is NOT a lossy format, but according to this here article IT IS. That`s also what I think. But then what you write makes no sense....<br />

Maybe I still don`t get your point, please help me again! :-(<br />

<br />

<br />

<br />

[Roseval]

<cite>You may not be aware of it, but since AIFF is not supported natively by Windows a similar conversion has to take place as any decompression conversion. The resources needed are about equal, with my personal conclusion that when you have the opinion to better not use FLAC (etc.) because of negatives like possible hiccups, one should better not use AIFF on a Windows PC just the same.</cite><br />

<br />

I don’t think this is correct.<br />

WMP supports AIF (the playback not the tags): http://support.microsoft.com/kb/316992<br />

Original AIF is of course big endian (power PC) so has to be converted on a Wintel.<br />

According to http://en.wikipedia.org/wiki/Audio_Interchange_File_Format Apple now has a little endian version of AIF (Intel Mac’s)<br />

<br />

[PeterSt]

Do you mean that Hydrogen article ?<br />

<br />

If so, what may confuse is that WMA is mentioned in the same sequence as MP3, Vorbis, but WMA is lossless and the others are not. Again compare with FLAC (because it is easy to see and use) :<br />

<br />

With FLAC you can state the compression level. 1 has less compression and the file will be "big", 8 is the highest compression level and the file will be "small". However, since at decompression all appears lossless, no matter which compression level is used, the bitrate will be 1411000 and it contains all that was in the original uncompressed file;<br />

Would it be possible to measure the bitrate for the from disk read data, you'd see that it would be lower at the highest compression level. This is logic because the file is the smallest (compared to the lowest compression level).<br />

<br />

Besides the above, if you'd look at certain passages in the FLAC file, you'd note that some passages (the more dynamic ones) need a higher bitrate, and others (more quite passages) just need a lower bitrate. The latter just needs less bits to register all, because not so much is happening. Think like "it is quiet for 5 seconds, so just register 'quitet' and 'that takes 5 seonds'. Could be done in a few bytes, while normally 5 seconds would take 5 x 176400 bytes.<br />

<br />

This is totally unrelated to variable bit rate means that look at the frequency, and state that when the frequency is under 11025 Hz, a sample rate of 22050 would be sufficient (while a frequency of 22050 requires a sample rate of 44100). Thus, in this case passages without the higher frequencies use half of the sample rate and it saves half of the space in that case.<br />

In theory this can result in lossless data (WMA). In practice nearly each sound goes way up to 22050 Hz because of its harmonics (only a pure sine of, say, 1000 Hz will just stay there), so longer passages going to 11025 Hz only (this is just a taken figure of course) don't exist. So what happens with lossy VBR (MP3) in practice is that analysis could show that reletively few high frequencies are there, and they are just taken out. For example, MP3 at 192Kbps (VBR or CBR) doesn't contain any information above 16KHz. It's just taken out.<br />

A 256Kbs MP3 does go to 22050 Hz, but in the higher frequencies or in the lower, not all data is taken, with smart presumptions on what we can hear and what not.<br />

<br />

In either case the difference between VBR and CBR is that VBR will require less data in the compressed file for the same quality as CBR would.<br />

All 'n all both can be used in a lossless format as well as in a lossy format.<br />

<br />

Maybe I should take back that "VBR" denotes lossy. As long as you know that it sure does not denote lossless.<br />

<br />

The main point remains that for a lossless file (WMA) you would not be interested in the fact that VBR is used. This is the same as that you are not interested in the compression level of a FLAC file. Note though that both (WMA/FLAC) are complete different means of "compression". FLAC does not make use of anything such as VBR (at least as far as I know), while WMA does. Yes or not, the quality stays the same.<br />

With lossy formats though, the quality with VBR vs. CBR (yes, now suddenly CBR is in the equation) sure does matter. With both means you can make an MP3 file, and while with CBR you can ask for 256Kbs with VBR you can ask for an estimation of the same, but the VBR will sure be of better quality at that same file size (that's what it comes down to).<br />

<br />

I can't do any better ... :-)<br />

Peter

[PeterSt]

Any player might support big Endian AIF, but natively Windows PCM is Little Endian.<br />

If a player doesn't do anyhthing about it (including WMP) it won't play ...<br />

<br />

If Apple now supports a Little Endian version of AIF, that would be COMPLETELY stupid. No two formats should have the same extension.<br />

If I am correct, it would need renaming to WAV to play on a Windows machine. Maybe WMP will dig it, but I sure won't (bother). To be precise, one week ago that Little Endian AIF would play on XXHighEnd. Today it doesn't anymore because AIF just is officially supported now. It will convert to Little Endian, and a mess will be the result (but I guess the header can't be read to start with).<br />

<br />

However ... maybe only half of what we both derive from that wiki topic is true, but nevertheles :<br />

<br />

<cite>Note: As of Mac OS X version 10.4.9, the system will sometimes incorrectly displays the AIFC icon for files with the .aif extension, whether or not the actual file format is AIFF or AIFF-C. This can be verified by opening the files in hex editor and checking the FORM chunk's form type. This can sometimes happen when exporting files from QuickTime, and frequently happens when sending and receiving files between Windows and Mac computers or extracting files from an archive.</cite><br />

<br />

Well, who cares. Right ?<br />

Prrrrt<br />

<br />

[haiku1]

Now I`ve got it.<br />

Thank you very much! :-)<br />

<br />

Kind Regards

[Idle]

I'm also using AIFF for my CAS. However, I have difficulties in converting AIFF into AAC for my portable MP3 device. Now I can only convert AIFF to WAV first and then convert WAV to AAC. What a trouble. Does anyone know some software can perform such direct conversion (AIFF to AAC)? Thanks in advance.

[silverarrows]

I had my entire library in FLAC and MP3 (320) and when playing through my Squeezebox 3 I could not tell the difference. Overall, I felt something was lacking and despite everything that I read about FLAC I was disappointed. I thought maybe I need to get a better power supply or try something else. My SB3 is being fed through coax into a separate DAC. I then ripped a few CDs into AIFF which was a mission on it's own. I used iTunes but it's just so slow. Foobar worked well but doesn't support tagging. I tried dbpoweramp and the files would not play. I tried EAC with an external iTunes rip and the sound was too low. So back to iTunes it was. After ripping a few CDs in AIFF, now I can't tell the difference between the CD and the rip. I even converted some of my FLACs to AIFF and I'm totally amazed at the difference. So now I'm trying to decide between WAV and AIFF. Long term I don't know if I'll be using a PC or Mac but seems AIFF will have the best in terms of options. My SB3 also seems to read the tags as well. Either way, I'm totally sold on uncompressed.

[markr]

To convert AIF(F) to AAC: - use iTunes!<br />

<br />

markr<br />

Good stuff connected to more Good stuff recording and Playing GREAT stuff!

[Superdad]

Peter:<br />

You were going along fine and helping clear things up for the noobs until you stated that WMA is a lossless format. By far the most common usage of WMA is very much a lossy format (see: http://en.wikipedia.org/wiki/Windows_Media_Audio). WMA Lossless is lesser used variant and not what is typically distributed.<br />

Alex

[firefly0071]

The differences in file formats must confuse lay people.<br />

<br />

It's like where hifi was in the sixities with component hifi coming onto the market in different product types.<br />

<br />

The file format must deter some people getting into digital music.

[The Computer Audiophile]

Hi firefly - I agree 100%.

[PeterSt]

<cite>You were going along fine and helping clear things up for the noobs until you stated that WMA is a lossless format. By far the most common usage of WMA is very much a lossy format (see: http://en.wikipedia.org/wiki/Windows_Media_Audio). WMA Lossless is lesser used variant and not what is typically distributed.</cite><br />

<br />

Yes, I f*'d up on this one.<br />

Luckily it doesn't matter much to the story and outlay because WMA Lossless indeed uses a variable bit rate.<br />

It explains though why it was in that list of "MP3 alikes".<br />

I am sorry.<br />

<br />

Btw, it can't harm to read that link. It's a little longish, but it may make clear better what this is all about in a nice natural way.

[Roseval]

<cite>It's like where hifi was in the sixities with component hifi coming onto the market in different product types.</cite><br />

<br />

There is a nice topic about this on Hydrogenaudio.<br />

The question was if there is a format you can rip to regardless of the OS and media player you are going to use. If both play back and tagging must be supported, the answer is an obvious no. <br />

[firefly0071]

Computer audio will push CD based systems to a minority market for sure in the future.<br />

<br />

But I don't know what epople will be using in 5 or 10 years time.<br />

<br />

USB 2 will be USB3 and there is a new firewire version coming out.<br />

<br />

So today's equipment and file formats may be junk in the future.<br />

<br />

A CD player 20 years ago is much more lilely to becompatible with todays systems than today's file format with something in 10 years time.

[The Computer Audiophile]

Hi firefly - I agree with you for the most part. This line of thought is also why I recommend AIFF. It has been around since 1988.<br />

<br />

<i>"WAV was developed for Windows 3.1 around 1991 by IBM and Microsoft. AIFF was developed for the most part by Apple in 1988."</i><br />

<br />

<br />

[flee]

First, thank you Chris for an informative and very timely (to me) discussion. I am using AIFF for my cd rips after listening to the various other options. I sadly had ripped about 500 cd's in mp3 quality before I realized the error of my ways!<br />

<br />

Lee Stone

[Roseval]

Most of us rip our collection twice.<br />

The first time because we rip it<br />

The second because we know how to do it

[Eesau55]

Hi,<br />

<br />

with Linux, FLAC is really the only way to go.<br />

<br />

Esa

[kevinfisher]

I was on the HDtracks store and was surprised to see they only offer flac in their hi res store. I am going to do everything I can in AIFF, but was wondering if there is something about flac that lends itself to higher res 96khz/24bit other than file size?

[The Computer Audiophile]

Hi Kevin - I'm not speaking for HDtracks, but I believe it offers high resolution material in FLAC strictly for the file size benefit.

[hubsand]

I'm surprised that only one or two comments in this thread have highlighted the all-important differences in playback quality between lossless formats - the discussion has proceeded as though we're talking about archiving digital photos. iTunes, for instance, makes quite clear the audible difference between Apple Lossless (ALAC) and AIFF. <br />

<br />

We're quick to damn compressed formats such as MP3 - and rightly, because they once for all time degrade the file - but the attractive reduction in file size offered by FLAC and ALAC also comes at a price: the on-the-fly decompression audibly exacerbates jitter. Just like notes, the gap between the zeros and ones matters, too.<br />

<br />

All of which points to AIFF as the only suitable format for playback and archival. FLAC and ALAC are fine for archival, but compromised for playback.