 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
87.228.167.35
In Reply to: RE: Best format to store music files posted by Dawnrazor on June 09, 2007 at 22:31:29
I am suspicious of FLAC even at 24/96 played backl on Foobar. HF seems to have an unwelcome emphasis.
Any form of compression entails the use of a statistical model on how to shoten certain types of audio informtaion, and an algorithm ecexecuted to recover the information for playback.
Computer software written to execute these are NOT tranparent as to what models and assumptions/truncations are used. If you absolutely believe in 'bits are bits' and programs comaaring them are infallible in terms of reassembling the audio string, then you will say that there cannot be a difference in sound quality.
I don't and believe I can hear differences. So I use wav at original resolution. HD space is cheap and fancy catalogging is not necessary. You cannot do this with a CD, so why on a computer?
Follow Ups:
Hey F,
When you say that you use the original resolution (16/44) is this because your DCS upsamples, or is this an indictment of upsampling in general???
> I am suspicious of FLAC even at 24/96 played backl on Foobar. HF seems to have an unwelcome emphasis.
> Any form of compression entails the use of a statistical model on how to shoten certain types of audio informtaion, and an algorithm ecexecuted to recover the information for playback.
FLAC compressor is lossless, it means that converting a wav file into a flac and bringing it back to wav will give you a bitwise exact copy of the original wav file (i've tested it today, it gives back a wav file with the exact CRC checksum), so i can't figure out how a lossless codec could even alter emphasis or dynamic response.
Lossless codecs (such as FLAC) absolutely don't use statistical or psychoacoustic models to achieve compression (rather than lossy such as mp3, wma ...). The work on the raw byte data of wav files in a similar way of zip(or similar) compressions.
Today i've spent some time to write down an article on my site. It contains also a guide to configure EAC to directly rip cd into flac files.
I hope it would be useful :)
http://www.primianotucci.com/go/flac
Primiano Tucci
(yep, I'm the son of the thread poster, of course ;-) )
You need to read up on FLAC!
its a pretty simple test
FLAC compressor is lossless, it means that converting a wav file into a flac and bringing it back to wav will give you a bitwise exact copy of the original wav file (i've tested it today, it gives back a wav file with the exact CRC checksum), so i can't figure out how a lossless codec could even alter emphasis or dynamic response.
I don't think that anyone is arguing that Flac isn't lossless, just that the process of "unzipping it" during playback can affect the sound. Several posters have heard differences between identical files.
This all may depend on the computer involved and extra things going on like use of crossovers, room correction, upsampling.
Do a search for posts from Christine Tham...she has some posts that give some plausible explanations as to why differences are heard.
I've just done a little test.
Using an intel P4 2.4 ghz Prescott cpu the flac decoder takes about 5 seconds to fully decode a 3 minutes flac file to wav.
Now, it actually depends how the player works... it could decode the flac to wav, and then play the wav file, or it could decode and play the flac wav on the fly (if you use the directshow filter the decode and play process is surely done on the fly).
However, doing a simple calculation based on empiric tests, it seems that the flac decoding process takes about 3% of cpu bandwidth (5s to decode 180s of data= 2,7% of real time cpu usage).
Sincerelly I find really hard to believe that an additional 3% could affect resulting sound
One of the great aspects of digital computating is that the response you get is discrete: you get the output waveform,or you lack it, but you can't get a degraded waferorm.
So, in my point of view, the only way (or better the main way) to compromise sound quality is falling in underrun of sound buffers (so you should hears gaps in sound).
In order to have a buffer underrun the cpu must be under severe load averages (and i don't think it's the case of flac decoding).
Howver this argument is really interesting me... as i'll get a bounch of free time i'm curious to try a file by file "Wav vs flac human test" ...i'll let you know my subjective impressions ;-)
Primiano Tucci
What you have done has no bearing on the sound or the accurracy of model or reconstruction
Using crossovers and upsampling on my 1ghz dedicated machine, I can tell you that on occasion I have done combinations particularly upsampling that have maxed out the processor, and made the sound behave like a 45 run at 33 or slower.
3% can make a difference for this pc, and from what you are saying, it is 3 % on your more than twice as fast pc or more than 6% on mine slow one.
let us know how your tests go. Others have heard a difference and I wish I had the spare time that you have to test this myself.
having gone with wav, it is pretty acedemic from me at least.
Hi fmak, I'm still trying to get my mind around this.
I take a face value that you can hear a difference. I really don't have a clear idea of how this all works, but you're thinking that it's the performance of the codec which is at fault? And since the dac is conversing with the codec algorithm, it is interacting with a procedure whose performance negatively impacts the resultant audio?
Do I have this right?
I don't know, but where there is no clarity in how things are done and I hear differences, I can only attribute this to the model or the programming.
See the link below for details on a small test I did in response to a post by Todd Krieger.
Alan
nt
| FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
||||||||||||||
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
