 |
|
You are not logged in! F.A.Q Log in Register |
 |
|
 |
 (nobody)...and 695 guests Last 5 registered Oplandisks nothingstar N_loop yipe foxtrotromeo Browse members... |
 |
Members 8025Messages 2614087 Today 0 Topics 127542 |
|
|||||||||
| |||||||||
|
|
so - most of you might have noticed that it seems that mp3 decoding off files affects alot of low frequencies and even in 320 kbps bitrate it still seems audible. now my opinion was the same until i checked it. i downloaded this analyzer from , feedbacked my audio card and created a file consisting of a few seconds of pink noise, a sine sweep from 10hz to 22000 hz, and a drumbeat i made. i'm not able to show you the results right now, i hope i can imageshack them later but i can assure you that the differences of power of the lowest frequencies in wav format and 320 kbps mp3 are nearly identical. that made me a little bit suspicious - i haven't found anything about that on google. but might it be that it's not really true that mp3 decoding affects those frequencies but that wav just "seems" better ? |
||||||||
|
|
|||||||||
| |||||||||
|
|
it's pretty standard for lots of mp3s to lop off frequencies below 10 or 20 hz completely, but those aren't audible anyway. any further loss of fidelity is prolly just due to a shitty encoder (not all mp3 encoders are created equal). if you really want to get to the bottom of this, you should compare one of the files which you think has diminished bass to a lossless version. |
||||||||
|
|
|||||||||
| |||||||||
|
|
that's why i was making the sine sweep from 10hz - it's below the audible range and i hoped the secret would be down there (something like the dithering noise which isn't audible too but people consider it sounding better. the problem is that i don't have any files with diminished bass. i did a very unfair test between wav and 96kbps mp3. the high-end was badly cut but till 10 hz the power didn't fell even 0,1 dB... |
||||||||
|
|
|||||||||
| |||||||||
|
|
if you don't have any of the files in question, why do you even care? what i'm saying is that in cases where you think there's been diminished bass due to mp3 compression it's prolly a bad encode, so doing your own encode isn't going to discover the "secret." |
||||||||
|
|
|||||||||
| |||||||||
|
|
now the thing is that that no matter how encoded, mp3 affects the lowest frequencies. or at least that's what it seems. it's not a matter of live and death in fact - i'm listening to almost only flacs right now and will still use mp3 even when i will be sure that it crops out the bass (at least it should so people will buy originals to listen to the full potential of a track) i'm simply curious that my tests didn't give any results and it's more important for me to find out if i made some mistake or the bass-cropping-mp3-decode is just a tale. |
||||||||
|
|
|||||||||
| |||||||||
|
|
the thing is also that it might simply be that the lack of bass in mp3 isn't due to cropping of lower frequencies but due to some other things. for example if you time stretch a bass sample it will also loose bass but you haven't altered the frequencie composition. |
||||||||
|
|
|||||||||
| |||||||||
|
|
you're only looking at amplitude. look at the spectrum of the sound and you'll see missing overtones, which contribute to a lot of the "feel" of bass. |
||||||||
|
|
|||||||||
| |||||||||
|
|
MP3s, if I remember correctly, first remove specific frequencies all across the spectrum, and then begin to top off highs and lows, and then at the very lowest-quality, basically run HPFs removing the top 5k and the lower 120 frequencies. Open up Pro Tools, and throw into it a lossless audio file, and then that same file decoded into 320k MP3. Polarity-reverse the MP3 file, and you'll hear all of the audio being taken out of the MP3 - the audio is made up of mostly midrange frequencies, because as long as the frequencies removed are exact and spread out, it really isn't noticeable to the human ear (and if you say otherwise, you're full of shit). |
||||||||
|
|
|||||||||
| |||||||||
|
|
trout: ah yeah i forgot about the polarity reversion - good idea in fact. phuker: how do you think i can measure that? a narrower octave split? i'm at the moment at 1/3 octave. |
||||||||
|
|
|||||||||
| |||||||||
|
|
seems like my fundamental knowledge in these things is still very low. i did what trout said with a track. there is in fact something left after doing this and it sounds like white noise. the problem is just that when decoding into mp3 it adds this little gap at the beginning and centering both tracks is very difficult so the noise that left might be amplified because of this. the problem is just that when reversing the pink noise or the sweep it sounds nearly identical... it seems i'm doing something wrong. |
||||||||
|
|
|||||||||
| |||||||||
|
|
Which MP3 encoder are you using? It should make a big difference. |
||||||||
|
|
|||||||||
| |||||||||
|
|
No one is completely full of shit after they have had a poop. But unfortunatly u cant just take the human ears frequency response alone into consideration, it affects timing as well. This will affect more live/acoustic music than most static electronic tracks. Timing that includes the hitting of a bass note or kick, and particularly when the 2 are in unison, the slight variation in time caused by removing freqs' and the re-building the the data can casue loss of energy/bass/definition of sounds. Other problems may include the decay of sounds/reverb tails. |
||||||||
|
|
|||||||||
| |||||||||
|
|
Nice point |
||||||||
|
|
|||||||||
| |||||||||
|
|
it's some "mp3-pro" thingie built into cool edit. i couldn't find any answers yet. |
||||||||
|
|
|||||||||
| |||||||||
|
|
Untilted in lossless sounds better than the mp3s i've had before imo. also i did that polarity thing with Ipacial section in Audacity comparing lossless with mp3/320 and the difference seems to be mostly in the higher frequency range... altering in the bass is not audible imo, but that's kinda logical innit imo cuz bass frequencies are pretty solid low resolution anyway. |
||||||||
|
|
|||||||||
| |||||||||
|
|
I cannot deny or confirm your claim, but I will say that in my previous post regarding polarity reversing one FLAC and one MP3 file, the time base would need to be the same or otherwise the example would not work. Surely, the timebase would have to be the same for both files, otherwise the polarity reversal would not cancel out the parts of the waveform that are unaffected, as demonstrated. |
||||||||
|
|
|||||||||
| |||||||||
|
|
yeah i put the lossless file into audacity and exported a mp3 from that. when i imported that file again, it was offset by a bunch of silence samples at the beginning, so i realligned it manually. If it was just one sample to the right or left, you would hear much more of the original song, but yes, maybe that was not accurate enough. dunno how to do it more accurately though... |
||||||||
|
|
|||||||||
| |||||||||
|
|
yeah i got something very similar sounding after doing that. lame can do gapless encoding http://www.geocities.com/nyaochi2000/lame/cuesheet/ didn't try it though |
||||||||
|
|
|||||||||
| |||||||||
|
|
if you're using mp3 it should be lame -v0 not to sound snarky, but if you have to make chartsngrafs of frequency response because you can't hear it it's probably not the biggest deal. as everyone else has already said the encoder you're using is gonna have a lot to do with that (most of them now have better perceptual coding than back in the days of everybody ripping britney spears and sharing it on napster) |
||||||||
|
|
|||||||||
| |||||||||
|
|
-v0 is transparent to me. i can't distinguish between lossless and -v0, so i've got most of my music in it. do an and find out if you can hear the difference between lossless and, say, -v0. ABX test is integrated in you'll have to check for ABX during install (if i remember correctly). |
||||||||
|
|
|||||||||
| |||||||||
|
|
important: "... a minimum of ten listening trials in each round of tests, as this allows a 95% level of confidence in results (...) also recommended that no more than 25 trials be performed, as listener fatigue can set in and skew test results". |
||||||||
|
|
|||||||||
|
Messageboard index
|
|||||||||
