Quality of LAME encoder vs FhG

W

WaveGuide

New member
Hi,
I was testing the difference between the LAME encoder and FhG. I got some odd results, so if anyone can make sense of this, please post your comments.

Anyway, I made a wave file which contained just pink noise. I then encoded at 320 with both LAME and FhG. Both gave rather good results, although there was a noticable difference. There was no real noticable drop-off until over 20KHz. However, whenever I tried the same thing with a music file, the LAME encoder does a much better job, while the FhG encoder drops off at about 16KHz. I can understand there being a difference between encoders, but why would the pink noise file come out so much better than actual music?
 
S

scarecrow

1
WaveGuide said:
Hi,
I was testing the difference between the LAME encoder and FhG. I got some odd results, so if anyone can make sense of this, please post your comments.

Anyway, I made a wave file which contained just pink noise. I then encoded at 320 with both LAME and FhG. Both gave rather good results, although there was a noticable difference. There was no real noticable drop-off until over 20KHz. However, whenever I tried the same thing with a music file, the LAME encoder does a much better job, while the FhG encoder drops off at about 16KHz. I can understand there being a difference between encoders, but why would the pink noise file come out so much better than actual music?
Click to expand...
Using pink noise for MP3 encoding is not an objective test. MP3 encoding algorithms are mainly based on advanced psycho-acoustic models, and quite obviously pink noise samples is not too suitable.
That said, Lame is VASTLY superior to FhG (although not speed-wise)- with good lame builds (under windows I prefer Dibrom's 3.90.2 and 3.90.3 builds, which you can find at www.hydrogenaudio.org - under Linux I toyed a lot with the switches and made my own 3.92 build ). I am pretty sure that a very high quality lame VBR encoding (say --alt-preset extreme) sounds much better that any FhG CBR encoding, while being much smaller in size.
But then it's no secret either that Ogg Vorbis is much better than .MP3, and Musepack (.mpc) encodings are close to perfection: average .mpc encodings using default settings like --quality 5 --xlevel, which sizewise are about the size of an MP3 encoded at 160K CBR sound transparent/identical to the ripped CD 99% of the times, even coming to expert ears and expensive hi-end stereo systems.
 
C

chronicking

New member
man i encode everything at 16oK/sec w/CBR. i'm sure im loosing quality but i really can't tell the diff. now i can with like 128k or 96k. it sounds muffled. i dunno maybe im gettin old but i thinx 160k is just fine. my question is, if you don't mind me askin, how do i distinguish which encoder my player(MMjb and WMP) will rip with? Where can i get the best encoders for what i need? and my WMP will sometimes not play certain mp3 files while MMjb will. i don't understand this. can someone post a link that might describe encoders and how to maximize their potentials? a good little free program i also use is CDEX. it lets you choose which encoder to use Lame, Fraunhoffer, etc.etc.....which arouses another question, if those encoders are available to CDEX, does that mean all my players and rippers have access to those encoders that come with CDEX?????? as you can tell, im abit confused about all this. if you can help, lemme know!!! lata!
 
W

WaveGuide

New member
the codecs you have available are shown in the windows control panel. look at the multimedia stuff. As for which codec is used by a particular program, that is another story. Some may just use the first one they find, while others may only use a certain one. If the program doesn't tell you, or offer a choice, I would not use it.
 
C

chronicking

New member
:confused:
on another note , i found something interesting this evening. someone on this thread posted earlier about the Kazaa reading the tag wrong.
here's what i found..........i was ripping(encoding, watever) at CBR(constant bit rate) 160/kps. i thought i couldn't tell the diff. well, many argued with me so i sat down in front of my hi fi at home ( has pc running to amp) and listened over and over to a STone Sour song that i ripped at CBR 160kps, CBR 192kps, & VBR 160-192kps. i could barely notice a diff if any at all. if i did notice a diff it was in the hi freq. range and not the lows (perhaps it was a better encoding of the voice. im unsure). anywayz that was enuff to convince me to use the VBR rate of 160-192kps. using that bit rate was just a bit bigger than the 160 CBR but not near as big as the 192 CBR. that tells me that 160 CBR is close but needs just a little more hard drive space for the complicated parts of the song. since there is no bit rate between 160 and 192 i reccomend using the VBR(variable bit rate 160-192). OK i know wat your thinxing "where the hell is this stoner going with this".
here's the kicker........Kazaa lite recognized all my id3v2 tags that were ripped at 160kps CBR but once i changed to the VBR of 160-192kps the tags were listed wrong thru Kazaa lite ( the my kazaa in the software engine) the times of the songs were wrong (usually longer than they actually are), the sizes of the files were listed wrong as well, and most annoying of all is that Kazaa lite listed the songs ripped at 128kps!!!!!! That's terrible cuz at this higher bit rate the song's quality is much better than at 128!!!! it makes me wonder how many songs ive past by in kazaa becuz of it listing it as 128kps.
Does anyone know how to remedy this. Now MusicMatchJukeBox read the tags correctly. Window Media Player 9 did as well, but the mini Windows media Player 2 read the times incorrectly as well.
Is there some kind of decoder i need to install that is new enuff to read these VBR mp3's accuratley? or just deal with it........ i dunno , do you?



this is a post from another thread in another forum that i posted. kazaa i know is irreletive here i know, but im more concerned with these VBR mp3 encodings of mine. my Windows media Player 6.4.07 won't play half of them but MMJB will. is there a special decoder i need for all my media players to play these mp3's? im encoding mp3's with cdex at this VBR (160-192) i alwayz used to use Fraunhauffer encoders at 160 CBR. im wondering if this VBR is worth it. does anyone have any input on this subject......lemme know, thanx

update:: hahaha here's a link that helped me a bit
http://www.microsoft.com/windows/wi...decs/audio.aspx
 
W

WaveGuide

New member
OK, depending on your encoding software, it will sometimes do this. However, I always use CBR, so I have not done thorough testing to determine the root cause. I do know that windows' built-in media capability reads the time incorrectly with some files (at least on some windows versions). This is not the ID3 tags, but it is mearly a calculated number based on the bitrate, sample rate, filesize, etc. I know because I wrote a program to read this stuff. Although the time can be put in a tag, this would not be accurate if a file gets truncated. Programs that are designed to deal with the various encoding methods can accurately deturmin the length, as well as the bitrate, sample rate, etc.
 
You must log in or register to reply here.
Top