Head-Fi.org › Forums › Equipment Forums › Computer Audio › Lossy Formats Can Supposedly Be High-Resolution...?
New Posts  All Forums:Forum Nav:

Lossy Formats Can Supposedly Be High-Resolution...?

post #1 of 12
Thread Starter 

I found a very interesting link today:

 

http://marlene-d.blogspot.de/2013/11/mp3-and-other-hires-formats.html

 

I am highly skeptical of much of it, but it is an informative read nonetheless.

post #2 of 12

Interesting read

post #3 of 12
Quote:
Originally Posted by marlene-d 
Why the constant bit rate (CBR) of 320 kBit/s when Hydrogenaudio recommends variable bit rate (VBR) in order to save on storage space? Think about it: 60 minutes of music occupy 137 megabytes when encoded with 320 kBit/s CBR. For VBR with an average of 240 kBit/s these 60 minutes take roughly 108 megabytes. A difference of 29 megabyte. In 2013, two to three photos on a smartphone alone consume this. We have to be realistic here: it might have been an issue 10 or even 5 years ago but nowadays with an abundance of storage space anywhere, surely we can afford bigger files.

Same argument can be used for using lossless files like FLAC. Storage is so dirt cheap nowadays (i.e. <$100 for 2TB!), why even bother wasting time reading and writing articles about lossy vs lossless? We have to be realistic here! biggrin.gif Plus, as audio technology gets better and squeezes more and more performance out of digital files, who knows how lossy vs. lossless will compare in the future. Might as well get the closest format to the original recording, be it ripped FLAC from CD or a studio master from HDTracks.
post #4 of 12
Thread Starter 

Very true. The main advantage of compressing audio would be when you have a portable device that you want to jam your music collection into. Otherwise, you can just put it all on hard drives without worrying about storage space.

 

I find it intriguing that in your signature you have a link (which I am already familiar with) attempting to discredit 24-bit downloads, but in your post you recommend HDtracks. I wish that site told you whether what you are downloading is the actual studio master or something else entirely. I have a few 24-bit downloads from them and they sound very different to the CD versions...to the extent that I suspect they are from two different masters.


Edited by Music Alchemist - 1/5/14 at 1:46am
post #5 of 12
Quote:
Originally Posted by Music Alchemist View Post

I find it intriguing that in your signature you have a link (which I am already familiar with) attempting to discredit 24-bit downloads, but in your post you recommend HDtracks. I wish that site told you whether what you are downloading is the actual studio master or something else entirely. I have a few 24-bit downloads from them and they sound very different to the CD versions...to the extent that I suspect they are from two different masters.

I don't necessarily agree with everything in that article, but I still find it as a very informative read with many good points to think about. I value well-mastered high resolution recordings (there are not that many of them), but find 24/96 more than sufficient based on high resolution albums that I heard to date. In my testing with some really good recordings/masters like recent "Kind of Blue", I can hear an improvement by going from 16/44 to 24/96, but not from 24/96 to 24/192. If the original studio master happens to be available only in 24/192, I would rather not tinker with it or down sample it, and just leave it as is, even if it seems like a waste. Also, I did not recommend HDTracks, just used them as an example of where to get studio masters. With HDTracks, there is a love and hate relationship. They have some great albums that are well worth being in high resolution, but as you probably know, they also have albums with questionable origin based on spectrum plots/etc.
post #6 of 12
Thread Starter 
post #7 of 12
Quote:
Originally Posted by gevorg View Post


Same argument can be used for using lossless files like FLAC. Storage is so dirt cheap nowadays (i.e. <$100 for 2TB!), why even bother wasting time reading and writing articles about lossy vs lossless? We have to be realistic here! biggrin.gif Plus, as audio technology gets better and squeezes more and more performance out of digital files, who knows how lossy vs. lossless will compare in the future. Might as well get the closest format to the original recording, be it ripped FLAC from CD or a studio master from HDTracks.

 

Indeed. Though the size difference might for example allow 15 albums on a portable device (mp3 & co.) instead of 5 albums (FLAC). There are actually people (like me) who are interested in saving space as much as possible without lossing any quality.

post #8 of 12

More subjective uncontrolled and biased listening tests. That's what the world really needs!

 

DSD is a pretty bad audio format for numerous technical reasons.

post #9 of 12
Thread Starter 
Quote:
Originally Posted by Marlene View Post
 

 

Indeed. Though the size difference might for example allow 15 albums on a portable device (mp3 & co.) instead of 5 albums (FLAC). There are actually people (like me) who are interested in saving space as much as possible without lossing any quality.

 

Gah! You're the author of that blog. I can't remember where I found your link, but I have a lot of questions for you. For now, I will keep it simple.

 

Thousands of albums in lossy format can fit on a 160 GB iPod. If you are going to use lossy, it might as well be high quality lossy. That much is apparent.

 

I'm more interested in your claims about ripping CDs to lossy formats, but at higher bit depths, to supposedly increase the sound quality.

 

You mentioned that "floating point decoding has to be activated under advanced options" in dBpoweramp, so I'm not sure if you were aware of this:

 

http://www.hydrogenaudio.org/forums/index.php?showtopic=102511&start=0&p=843664&

 

Greynol (one of the most prolific posters on Hydrogenaudio) said: "The depth of an MP3-encoded file is 32-bit float."

 

Spoon (developer of dBpoweramp) said: "In dBpoweramp it reports what it decodes to, by default mp3 decode to 16 bit. You can change in dBpoweramp Configuration >> Codecs >> Advanced 'Mp3 decode to' 32 bit float, then dBpoweramp will report all your mp3s as 32 bit float..."

 

He also said "You would not want IEEE when ripping." here:

 

http://forum.dbpoweramp.com/showthread.php?24361-IEEE-32-bit-floating-point-when-ripping

 

The sample size (bit depth) that dBpoweramp's Audio Properties tab reports in all existing audio files' properties is merely dependent upon your configuration settings. In other words, it would appear that changing this setting would change nothing, as far as ripping CDs is concerned. This is true for both MP3 and AAC.

 

I tested this out further by ripping a track in dBpoweramp with four configuration settings:
MP3 (16-bit)
MP3 (32-bit float)
AAC (16-bit)
AAC (32-bit)

 

Then I ran a bit comparison using foobar2000. It says the audio data in the MP3 files is identical. Same goes for the AAC files.

 

Since these lossy files are bit identical, wouldn't that mean that the differences between a lossless and lossy CD rip would be the same, regardless of whether you selected 16-bit or 32-bit in the dBpoweramp Configuration?

 

What tangible benefits could be derived from your strategies if the resulting files are actually the same?

 

In a nutshell, I would like you to explain more clearly what you are saying about ripping CDs in lossy format with the seemingly-redundant 32-bit setting, especially taking into account what I mentioned in this post.


Edited by Music Alchemist - 1/5/14 at 8:48am
post #10 of 12

CD audio is 16-bit so it shouldn't make any difference if you encode that directly or go the 32-bit float route.

 

If you have 24-bit files then it usually will make a difference compared to feeding the encoder re-quantized 16-bit data, but this difference can be fairly small depending on the en/decoder.

 

edit: Just because an encoder can accept 32-bit data doesn't mean it will achieve over 16 bits of performance. Same goes for the decoder.


Edited by xnor - 1/5/14 at 9:05am
post #11 of 12
Quote:
Originally Posted by Music Alchemist View Post
 

I'm more interested in your claims about ripping CDs to lossy formats, but at higher bit depths, to supposedly increase the sound quality.

 

I´ve never said that one should rip CDs to lossy formats at higher bit-depths. That would be pointless.

 

What I said was that an mp3-file that was encoded using 16-bit material should be decoded to 32 bit float, no matter the bit-depth of the lossless file used for encoding. Therefore I agree with all the quotes you posted. Ripping to floating point is stupid since it only adds zeroes; it does not yield additional resolution.

 

I have to ask: did you understand my goal for the article? I wanted to stress out how important the correct decoding of lossy formats is and not that you should use floating point for CD ripping.

 

I want to make it clear: any lossy codec accepts any bit-depth, it doesn´t matter if it´s 16, 24 or 32. Internally all lossy codecs are floating point anyway which means their output (when decoding) doesn´t have a specific bit-depth unless it´s configured by the user.

post #12 of 12
Quote:
Originally Posted by xnor View Post
 

CD audio is 16-bit so it shouldn't make any difference if you encode that directly or go the 32-bit float route.

 

If you have 24-bit files then it usually will make a difference compared to feeding the encoder re-quantized 16-bit data, but this difference can be fairly small depending on the en/decoder.

 

edit: Just because an encoder can accept 32-bit data doesn't mean it will achieve over 16 bits of performance. Same goes for the decoder.

 

I´ve never said that you´ll achieve 32 bit performance by decoding an mp3-file that was encoded from a 16-bit file. The only thing I´ve said is that you´ll gain distortions that weren´t there before, distortions that can be avoided by 32 bit floating point decoding.

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Computer Audio
Head-Fi.org › Forums › Equipment Forums › Computer Audio › Lossy Formats Can Supposedly Be High-Resolution...?