[chewy4]

I've had a FLAC file with 99.7% compression before

. I think just using level 5.
 
"Magic Window" by Boards of Canada.
 
I don't think the computing difference is going to be very big for the different levels, but this isn't something I've tested. I just set it at default to be safe, and like skamp said it doesn't make that big of a size difference.

 

[skamp]

It depends on your usage pattern. If you encode to FLAC while ripping CDs, you won't lose any time using the highest compression setting. If you want to transcode an entire library in one run, using default (-5) might save you an hour or two (or more, depending on the size of your library) of encoding time.

 

[StewieMonSta]

A much higher quality than MP3's. A professional quality.

 

[Satellite_6]

Quote:

A much higher quality than MP3's. A professional quality.

Professional quality would be 24 bit, and there isn't much difference from mp3.
 
I use level 8 because I can, averages 70% of the original file size. 

 

[TubeStack]

... I doubt many would be able to notice a difference between FLAC and 320kb MP3...

If listening to a familiar song on a good system, the difference is very noticeable, especially in low end definition and punch/impact.

 

[chewy4]

Quote:

If listening to a familiar song on a good system, the difference is very noticeable, especially in low end definition and punch/impact.

The low end is the region the least effected. I don't think it's even touched in 320kbps LAME mp3 compression.

 

[TubeStack]

The low end is the region the least effected. I don't think it's even touched in 320kbps LAME mp3 compression.

Is that based on your own ears or reading things online? It's very apparent on a medium to large home loudspeaker system, in my experience.

 

[chewy4]

Quote:

Is that based on your own ears or reading things online? It's very apparent on a medium to large home loudspeaker system, in my experience.

It's based on several things. My own ears, my understanding of the compression algorithms and how digital audio data works, tests I have done...
 
Consider a bass sine wave - let's say 100Hz. It only takes 200 samples per second to accurately reproduce. That's not very much data.
 
Now take an upper treble sine wave - let's say 18kHz. That takes 36,000 samples per second to reproduce. A bit more data, to say the least.
 
Why on earth would a compression algorithm pick apart the bass when such a small amount of data is needed for it, when there are plenty of frequencies that most people can't even hear(or are hardly present in any music in the first place) that it can pick apart a lot more for very little, if any impact at all?
 
And this isn't just me speculating, this is the way it works. Tested and true.
Bass isn't touched. Some very difficult to hear harmonics on cymbals and some other percussion is.

 

[TubeStack]

Quote:
It's based on several things. My own ears, my understanding of the compression algorithms and how digital audio data works, tests I have done...

Consider a bass sine wave - let's say 100Hz. It only takes 200 samples per second to accurately reproduce. That's not very much data.

Now take an upper treble sine wave - let's say 18kHz. That takes 36,000 samples per second to reproduce. A bit more data, to say the least.

Why on earth would a compression algorithm pick apart the bass when such a small amount of data is needed for it, when there are plenty of frequencies that most people can't even hear(or are hardly present in any music in the first place) that it can pick apart a lot more for very little, if any impact at all?

And this isn't just me speculating, this is the way it works. Tested and true.
Bass isn't touched. Some very difficult to hear harmonics on cymbals and some other percussion is.

Interesting, although based on my own ears and experience, I can still hear (and primarily feel) a difference in the low end articulation and impact. Agree regarding cymbals, etc.