[ffdpmaggot]

In the new itunes, I found this new feature which allows one to convert from 256 kbps AAC to apple lossless files. I understand that just changing the file type wouldn't improve quality, but I know that there are certain algorhythms that can restore some lost detail by evaluating very faint sounds in the recording, and then adding some extra volume to them, or generally cleaning up the sound. So is itunes doing this? Or is it possible that itunes is now using a different audio codec as their standard, and itunes is really downloading replacement files? My final question is whether my ipod nano 4th generation can play lossless files back. I'm aware that my PMP isn't very good at all, but, the sonic signature seems to equal that of my PC sound card, and the noise floor is much quieter.

 

[khaos974]

There are softwares that claim to "improve" the sound of a lossy file, but iTunes is certainly not doing this. There is absolutely no point in converting am aac file into an ALAC one, it just takes more space for the same quality. Itunes is not redownloading the lossless versions of your files, it's just a basic conversion.
 
Furthermore, the softwares that claim to "improve" the sound sound of lossy files don't work really well, what they do is a combination of the job of a harmonic exciter, dynamic compressor/expander, parametric equalizer and probably with a limiter at the end. This is basically the tools of a mastering suite, except that depending on the type of music recorded, the compression and how the original mastering engineer mastered the track, you would need a different setting for every album (to somewhat add more fullness, warmth, restore the High frequencies of your track). There is no one "work for all" setting. I would hazard to say this kind of automated improvement software improves the sound of 10% of the lossy files, change the sound sig (neutrally as far as SQ is concerned) for 20% of them and degrades the sound quality of the rest of the lossy files.
 
Besides, 256 kbps AAC files is first rate quality for lossy compression, no need to improve anything.

 

[ffdpmaggot]

Well there is a need to some degree, I used to download albums from itunes (I use CDs now) and am looking into a moderately high end set-up, one which would sound best with lossless. I know that 256 AAC is the best lossy format, but it is still lossy and if I could, I would love to be able to improve the sound. As an unrelated question, I'd like to understand a bit more about how audio is encoded. Currently, I believe that kbps represents how many tones are played in a second, e.g. one kilobit. 320 kbps then means that there are 320 signals that are being sent out in a second. Each signal describes the sound that the headphones should be making at that point in time. So, if we are playing sweeping tones, kilobit 1 might be 300 Hz, and kilobit 2 might be 301 Hz. I thought that the supposed "decompressing" applications just filled in a different kilobit, say 1.1 at 300.5 Hz to increase detail in the sound. Is that something that is done/can be done if my interpretation of audio encoding is correct?

 

[khaos974]

Your interpretation of digital audio is quite wrong, the 320 kbps means 320 thousands (from the kilo) successive values of 1 or 0 per second. Frequencies belong to the time domain, so it's impossible to "encode"  a frequency in a single value, it's encoded in the succession of values.
 
To understand more about digital audio go here: http://en.wikipedia.org/wiki/Digital_audio, to specifically learn about how how CDs, wav and DVD-Audio are encoded, go here: http://en.wikipedia.org/wiki/Pulse-code_modulation
 
MP3, AAC and all lossy codec work by 2 principles, the first is  lossless compression, if the successive bits are 000000001111, instead of writing this, it writes 8*0 and 4*1, this is a very simplified analogy. the second is discarding inaudible data, for example, if a soft sound follows a loud sound very closely, the human ear will drown the soft sound into the loud one. The degree of inaudibility of th discarded information depends on the bitrate you chosse, the higher the bitrate, the less audible the discarded sounds will be.
 
for more info http://en.wikipedia.org/wiki/Audio_compression_%28data%29

 

[Bricolage]

Don't encode lossy to lossless, you won't gain anything. It's like scanning a printed paper and hoping it comes out as good as the original.

 

[ffdpmaggot]

Is there anyway to have it write "8x0" without throwing away soft sounds? That would certainly save space, but I don't see how it could reduce sound.

 

[dsf3g]

Furthermore, any software that could improve a 256k by converting to a higher bitrate could just as easily improve the sound on the fly while decoding and leave the source file at 256K to save space (sure the process might be too processor intensive to actually decode on the fly, but I'll bet there's a 99% chance that it's not too processor intensive for modern computing devices).

 

[silverxxx]

Quote:

Is there anyway to have it write "8x0" without throwing away soft sounds? That would certainly save space, but I don't see how it could reduce sound.

Well, if it didn't throw away soft sounds it wouldn't be lossy after all. To encode more information you would need higher bit rates,  and they would be close in size (or bitrate) to lossless files anyway, so it would not have any advantage over lossless files to start with (their advantage is size). On top of that they would still be lossy, since that is how the algorithm is designed to operate.

 

[khaos974]

Quote:

Is there anyway to have it write "8x0" without throwing away soft sounds? That would certainly save space, but I don't see how it could reduce sound.

 
Yes, it's called lossless encoding, the most popular codecs are FLAC on Windows/Linux and ALAC on Mac OSX.

 

[ffdpmaggot]

well couldn't some compression be a good thing? take, for example, this diagram
0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000010000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000011 00000000000000000000000000000000000000000000000000000000000000000000000000000000000
 
Those individual vibrations don't do anything, except make the file larger, could we get rid of those and plausibly clean up the recording a bit and reduce the noise floor?

 

[silverxxx]

Quote:

well couldn't some compression be a good thing? take, for example, this diagram
0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000010000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000011 00000000000000000000000000000000000000000000000000000000000000000000000000000000000
 
Those individual vibrations don't do anything, except make the file larger, could we get rid of those and plausibly clean up the recording a bit and reduce the noise floor?

Yes, that is why FLAC's compress at all. However, you cant clean or improve anything because the codec does not know what is what. If it improved something in one song it might remove something the artist included in another. Even noise may be added to a song for some purpose (and is), the codec's job is to reproduce the recording as it was. Otherwise it would be a DSP.