LAME vs Fraunhofer

[Brian]

Any views on the respective merits of LAME (3.97) and Fraunhofer (4.03) for encoding mp3s, please? I've just upgraded to the Reference edition of dBpoweramp, so I now have the choice. Many thanks.

[Dotpitch]

LAME is widely used and is very good. But the Fraunhofer Gesellschaft have done years of tuning and work professionally on optimizing their mp3 libraries.
I've never tried Fraunhofer, actually, and all Google could give were some ancient tests (back from LAME 3.90 and Radium). You can test them using an original CD track you know well, encoding it at various bitrates and comparing the three spectra. I doubt there's any deviation in quality when you use high bitrates, so I'd go for the faster one.

[kenshin1101]

Lame : best quality + Low Speed
Fraunhofer : Fast Speed but quality < Lame
i choose the quality

[Brian]

Thanks for the replies.

kenshin1101: Can you say in what ways the quality of Fraunhofer is inferior to that of LAME, please?

[kenshin1101]

currently Lame is the best VBR MP3 encoder.
You can try it your self

[Dotpitch]

Robertos test of MP3 at 128 kpbs is from 2004, but it shows that FhG is just a bit behind LAME. This thread on HA has a lot of not-to-the-point conversations, but it has some messages from people that are content with the FhG encoder. If you're going for high quality MP3, I'd still go for LAME VBR. From that thread:

LAME has a much better proven track record, since a LOT of public testing has been done on it. It is possible that other encoders could be better, but they don't have the testing behind them than LAME does.

[Brian]

Dotpitch - thanks for the useful pointers.

The physical test by Roberto looks worthwhile, although of course a lot can change with software in four years.

The stuff on the HA forum all seems rather anecdotal (not to mention the discourtesy shown by forum regulars towards a new person asking perfectly valid questions).

So far, I'm pleased with the results from Fraunhofer at high bit rates, although I wouldn't claim that they are superior to LAME results. I might just encode my music collection to Fraunhofer (it would only take a couple of hours, and I have plenty of space on the drive), and see how I get on with it over a period of time.

[Sam_Zen]

Besides the discussion about quality I like to introduce another aspect here:
Afaik LAME is open source, while the Fraunhofer institute announced a while ago to have plans to introduce a copyright-payment for every MP3 used on the web.

[Dotpitch]

Afaik LAME is open source, while the Fraunhofer institute announced a while ago to have plans to introduce a copyright-payment for every MP3 used on the web.

Fraunhofer and Thomson have a patent portfolio for different aspects of MP3 encoding and decoding. That LAME is open source, does not mean you bypass the patents in any way. Licensing is done from mp3licensing.com, and private use is free.

[...], no license is needed for private, non-commercial activities (e.g., home-entertainment, receiving broadcasts and creating a personal music library), not generating revenue or other consideration of any kind or for entities with associated annual gross revenue less than US$ 100 000.00.

[Pike84]

LAME vs. Fraunhofer in 2008? Heh, I didn't even know that Fraunhofer's encoder was still being developed .

I'd go LAME VBR all the way without thinking twice about it. The latest version is 3.98, btw. You can read about it here.

[gigalot]

I have tested 2 files in Adobe Audition 3.0 frequency spectrum. One was Lame 3.98
--preset xtreme (highest quality VBR)
and the second Fhr mp3sEnc_Win32_FCRv14_20070711 with
-if Track01.wav -of Track01.mp3 -br 0 -m 1 -q 1 -vbri (highest quality VBR)
Files has about the same size.
But the spectrum show me that Lame max frequency was about 19000 hz with some unnormal lines wasn`t in original WAV and Fhr Mp3sEncoder max frequency was 22000 without any bad lines. Lame has a little more high freq in sound, and for me Fhraunhofer sound was still tha same as WAV and better than LAME!

[Pike84]

I wouldn't think too much about any spectrum analyzers, or anything like that - the most important thing is what you hear. Myself, I can't even hear a difference between the original and a 128k/s CBR encode in most cases, let alone a modern VBR encode.

Also, 19000Hz is such a high frequency, that you'd have to be some kind of a batman to hear sounds of that range . Besides, limiting the frequency frees extra bytes for the encoder to use on some more difficult passages to improve the sound.

If you truly hear a difference in the sound in general, not just single artifacts, at such high bitrates with VBR encoding, then it's very likely placebo. To confirm there's a real audible difference between these encoders, you could try ABX double blind testing method with some short music samples.

But it doesn't really matter.. By all means, use whatever encoder you think is best .

[gigalot]

 Some of Yung people 20-25 years old can hear 25000 hz and 1-2 from 1000 about 30000 hz in darkness! But most people at 50-70 years old can hear only 16000 as maximum.
 
Sometimes music instruments has a very high sound attack and it be better on WAV or 320 cbr MP3. It isn`t always placebo.
For most people - yes, but not at all. For these guys WAV is better than 320kbit/sec mp3 far from 128 kbit/sec. And sometimes you can hear sound artifacts on 128. Anyway, Lame`s quality on 128 cbr is very good and useful. If you have best speakers 192kbit/sec may be better choise

[gigalot]

I desided to wite some interesting for Fraunhofer funs because it is latest Frh stereo & 6chanel  command line free mp3 encoder and has advantages:

o Encoding a multichannel WAV file larger than 4GB:
mp3sEncoder.exe -if 6channel.wav -of output.mp3 -br 320000 -eof
o Encoding six mono WAV files:
mp3sEncoder.exe -if infile_L.wav -of output.mp3 -br 320000 -monofiles
o Encoding a six channel interleaved PCM file sampled at 48Khz:
mp3sEncoder.exe -if 6channel.raw -of output.mp3 -br 320000 -raw -sr 48000 -c 6

* Once you have your mp3 Surround file, simply use VirtualDub or VirtualDubMod to mux the audio with a seperately prepared DivX video file. To these applications the mp3 Surround file appears to be a regular CBR Stereo MP3 file, but DivX Player will detect the additional channels during playback.

I newer see any special GUI or Front-End for this ecoder but i found that Exact Audio Copy, which is best CD ripper is also nice GUI for it! You must choose "MP3: Fraunhofer Mp3Enc" in external encoding parameters after you put Mp3sEncoser.exe to EAC folder. Disadvantage: only cbr encoding from 20kbit/sec to 256kbit/sec is available there, but it is good for MP3 mass producing.
VBR is working from command line...not works in EAC.

Free download Mp3sEncoder.exe is here: http://www.all4mp3.com/tools/sw_fhg_cl.html

[Pike84]

Some of Yung people 20-25 years old can hear 25000 hz and 1-2 from 1000 about 30000 hz in darkness! But most people at 50-70 years old can hear only 16000 as maximum.

Actually, most adults can't hear sounds above 16khz - you don't have to be 50 years old. Don't know about these special cases.

Anyway, Lame`s quality on 128 cbr is very good and useful. If you have best speakers 192kbit/sec may be better choise

Remember that VBR is always superior to CBR at similar bitrates.

[Zarggg]

LAME (and any lossless algorithm, for that matter) can produce artifacts using VBR or CBR, regardless of bitrate. The only way to determine what is "best" is to do blind ABX testing from Lossless sources.

[gigalot]

Yes, VBR is better than CBR. But CBR files is MP3 standard and sometimes standalone players has problems with VBR stream. VBR files has a special tag with information of file size.
Older players can`t read it and cause problems with rewind and it can cut last samples or first bytes. The next disdvantage is that some players cut high frequences we love after 16000, so the player for VBR must be very good!
Old Winamp 2.22, 2.70 has no problems and Foobar 2000 is very good too. Use VBR for fun!

[gigalot]

I put my music fragment encoded by Fraunhofer Mp3sEncoder VBR mode and audio spectrums here:
http://rapidshare.com/files/129921243/Test.7z.html

What do you think about it friends?

[gigalot]

Musepac .MPC which is pretty good music format often seems better than .MP3 has a frequency limit 19000 hz in its --Xtreme encoding according with Lame 3.89 (but old Lame 3.90.3 people like has 20000 hz limit at --Alt_preset xtreme) and above 20000 hz at --insane encoding files.

If you can hear such high frequency you can try --xtreme or better settings.

[Zarggg]

Yes, VBR is better than CBR. But CBR files is MP3 standard and sometimes standalone players has problems with VBR stream. VBR files has a special tag with information of file size.

Actually, VBR is now the default and preferred mode for LAME.