hi,

i'm new at this forum.

i read the postings about lossy compression formats since a few weeks. for me as musican the mpc format sounds better than all other lossy formats. i have make any listening tests with mp3 --alt-preset standard and mpc -insane. for me the mpc's encodet with -insane sounds very good. the frequencies od these files are fine. the stereo room are fine. but the compression goes away. i think that mpc is too much transparent. the files sounds "thin" and "unreal" for me.

does anyone hear the same effect?

Dezibel

Nope.

You might want to try ABX testing. You may be hearing what you want to hear...

I used to have the URL to the ABX utility, but I no longer have it. Maybe one of the others can post it? And in case you're wondering, the ABX utility is a blind test so you don't know which file you're being presented with. You must then pick out which is which a number of times and it'll tell you your accuracy.

Others can probably elaborate much better than I can

i'm used winamp with random function to make my hearing tests used AKG k240df headphones to be sure no room and speaker quality things would change the sound.

Dezibel

Here:
www.pcabx.com

Hello Dezibel!

To avoid misunderstandings:

Generally, when discussing lossy compression, referring to a setting, codec etc. as 'transparent' means that the end result is more or less indistinguishable from the original. In other words, pretty damn good.

/ Uosdwis

my english is not the best. with "transparent" i mean that the sound is "synthetic".

Dezibel

Well, according to some people MPC also sounds "hollow", "bassless", "too clear", "too bright", "noisy", "put your expression here..", etc. etc. Same goes to every other codec as well, there are dozens of expressions for every codec often totally in contradiction.

Please provide your ABX results, and make sure you don't use any clipping prevention,replaygain,WinAmp EQ, DSP plugins, etc. etc. when doing the comparison.

Ps. Sorry, I accidentally first answered into your message, that's why it shows it's edited by me.. it's 6am for me and I pressed the wrong button.. sorry

*lol* it is 5am for me ;-) but it is ok

Dezibel

i think that the developers of the lossy audio codecs are trained with hearing tests and they are now what kind of artifacts they should search in these encodet files. i see music as not an technically thing. for me music lives, and i can hear if the guitar player lost sleep last night ;-)

i don't understand that no one hear the same effect using mpc.

Dezibel

The reason is most likely that you are just imagining you are hearing differences.

When I developed the GTuned encoders, I gave CD's with the several tracks,
both not compressed, with LAME --r3mix and with Ogg GT2 to a few friends
that are musicians.

Neither of them succeeded in hearing _anything_ wrong with _any_ of the tracks.

The bottom line is, being a musician won't allow you to distinguish a codec
easier, far from it.

The errors a lossy codec makes are not ones that affect the feel of the music.
That is something that lives mostly in your mind, and the suggestion that
you are listening to a lossy codec will ruin that feeling.

--
GCP

for me an mp3 encodet with --alt-preset standard sounds pretty cool. not 100% of the original, but cool. if i encode that file with --alt-preset standard -k, it sounds much more "real" for me.

my expirience is, that ultra high frequencies [20 - 22kHz] are very important for the "live-feeling" of an song. i've tested my ears with an 0 - 22kHz sweep tone and find out that i can hear up to 16kHz. although i can't hear an 20kHz tone, this frequencies are important for the life feeling of an song.

for me the mpc codec brings the best results of all lossy codecs because it is the only codec that saves "all frequencies" [-insane preset]. but i think that to much "compression" goes away. ...hope that anyone understand my point

Dezibel

Well, many people have said the same thing as you, including me. But people who can hear up to 20-21kHz sine tone, can't normally distiguish 19kHz lowpassed music from full bandwidth in blind ABX tests, even though without blind testing when knowing the frequency response, the full bandwidth "feels" better.

Unfortunately these "feelings" have nothing to do with the reality, unless you can prove it with blind test results.

Isn't the "feel" of the music paramount to the listening experience? MPC is virtually artifact-free...that is the main reason that I use it. I find artifacts really degrade from the intended experience. However, there is more to music than just sounds.

I don't listen to music to hear the sound. I listen to music for the "feeling" that it elicits. Whether it is proveable or not that MPC sounds different, the psychological response is there that it "feels" different. And the psychological response is all that matters as far as I'm concerned. Music is just noise; the mind makes it "real."

After using Dibrom's sweep.wav file, I was bitterly disappointed to learn that I cannot hear anything over 17KHz. I burned the track to CD-R and tried the test on multiple playback devices to make sure my sound card/Windows settings weren't to blame, but I found the same results.

It's strange because I'll listen to the sweep and then all of a sudden the sound output just drops off...it's not a gradual decline. The file hits the 3.5s mark and bam, it just goes quiet. I did test my hearing threshold using Kevin Donohue tone sequence test and didn't find any problems in the 4KHz-16KHz segment and discovered I am rather sensitive to frequencies below 100Hz.

As a result, I've added the --bw 17000 switch to my MPC command lines and haven't yet noticed any high-frequency loss compared with the original WAVs.

this may not be the best idea, as there have been studies showing that even frequencies you cannot hear affect the ones you can. you will understand what i mean if you know about the idea that when 2 frequencies collide, they produce a third frequency.

personally, i suggest the default audio bandwidth of 22050hz so that all of the info that is in the music from the CD is considered for encoding. basically by setting the bandwidth lower, you are telling the encoder not to consider frequencies above 17000khz when it decides what samples to encode and which to not encode (if this is not true, someone please correct me)


ok as for the "tinny" sound, i dont think this is true. although many others have voiced this concern.

because of the subband nature of MPC, it is natural for it to sound crisper and clearer than transform codecs (again if i'm wrong please correct me). overally the songs and sounds soud clearer.

basically, when people are used to listening to MP3 and other lesser quality files, i think it may be natural to think something like MPC sounds "too clear" or "too defined"

[b]Dezibel when you say "compression" what do you mean?

do you mean the volume of the file?

please elaborate.

All I know is that I cannot reliably tell the difference between a music track with a 17KHz lowpass and one that is full bandwidth. Test tones, maybe, but not on "normal" music. If frequencies above my hearing threshold have an effect on those below, I can't find the evidence personally. And since using a lowpass often spells attractive reductions in bitrate, I think it's advisable to use it, especially for MP3 with its inefficient sfb21.

About the vague "thin" sounding of mpc. I'd like to quote Frank Klemm from what he wrote here about SV8:



Isn't that exactly what we are talking about? Sure, I also can't distinguish that peculiarity of mpc in ABX. But it's there, it is a feeling, not provable. Couldn't it be that it relates exactly to what Frank states here? Just a small overemphasis of high frequencies?

Northpack

This does not work as you think it does. The only way it is possible to get this effect is by additional intermodulation within the speakers themselves.

(I wanted to give more detail on this, but I can't seen to find the relevant threads.)

--
GCP

AFAIK it's a difference of 1dB or less. Considering this is considered to be on the edge of perception for the full bandwidth, it seems extremely doubtfull it is audible within the frequency range where our hearing is worst.

--
GCP

<quote>
Dezibel when you say "compression" what do you mean?

do you mean the volume of the file?

please elaborate.
</quote>

compression says that the song is compressed when i record an song in the studio i can't master the song with the full dynamically range. i have to "compress" the song before i can burn them to cd or encode as mp3. if you use the full dynamically range the song sounds not good for people they don't musicans. you can normalize an song in 2 ways. first you can maximize the peak value. that means you let the highest peak set on 0db. the second way is to maximizee by RMS level. that means that the subjective db level will be checked and the song will be normalized by compressing. if you look on an compressed song in an wave editor you will see that the highest peaks will cut off.

if you don't now how about i'm talking load an "oldie" in wave editor [soundforge, wavelab, cooledit...] and look at the highest peaks in the song. than load an newer dance, techno or hiphop song and watch how the highest peaks where cut off.

this is the reason why techno and hiphop listeners like to hear vinyl. the vinyls have more compression than a cd. songs with synthetic sounds within, like techno, or hiphop became from the vinyl an "warm", "real" feeling.

@Nortpack:

...it's not an treble problem!

<quote>
ok as for the "tinny" sound, i dont think this is true. although many others have voiced this concern.

because of the subband nature of MPC, it is natural for it to sound crisper and clearer than transform codecs (again if i'm wrong please correct me). overally the songs and sounds soud clearer.

basically, when people are used to listening to MP3 and other lesser quality files, i think it may be natural to think something like MPC sounds "too clear" or "too defined"
</quote>

i think this would be the reason i have to check them out later. but i haven't much time at the moment to work on music

Dezibel

how can i "quote" in this forum?

Dezibel

Use (without spaces) [ Q U O T E ] [ B ]
and close with [ / B ] [ / Q U O T E ]

--
GCP

None of the lossy codecs you talked about does anything that changes the volume or compresses the dynamic range of the music.

This pretty much confirms my initial guess that the problems you are hearing exist solely in your mind.

Please use proper blind testing prodecures and ABX before claiming to hear problems, thank you.

--
GCP

Press the quote button related to the poster you want to quote..
Then for more specific or additional quoting: [ quote ] blablaa [ /quote ] (without spaces)

i will test this abx tool if have a little bit more time. thx for help

Dezibel

One useful link about that: http://www.digido.com/

As for the problem you have with mpc, yeah: ABX it

I understand that this is an old thread but those of you who think that quality is not up to par with MPC you need to get ahold of Rush 2112 listen to the Overture encoded with --insane. The bass felt like someone hit me in the chest with a hammer.
My equipment
Sansui G9000 Reciever
Marantz Imperial 8 Speakers
Pioneer PD-f905 CD Player