i want to do some tests with some encoders and i need the fhg mp3 encoder.
i think there are quite some, so what are the differences between acm producer pro, or fastenc etc.

do they cost something or can you download them freely ?
where to get them ?

The latest version of Fhg is in Cool Edit and Nero Wave Editor.
But in Cool Edit you can change the lowpass and other settings when in Nero Wave the options are less.

I think that FhG still have the mp3enc demo available for download.
Mp3Enc is slow, but it's a good encoder.
You can use it for testing, as the demo is limited only to 30s encodings. That should be enough for test tracks... (that is what you asked, didn't you?)

I believe a full version of FhG MP3 and MP3Pro encoder is included into the free version of the MusicMatch Jukebox 7.5. You can select different "quality" levels and lowpass filter settings, etc

Using the latest [fastenc] cooledit filter with EAC might be smarter than installing any bloatware: all settings are available plus you can rip and encode on the fly. The filter is very small too (less than 100K as far as I remember).

How do you get to use the cool edit fhg filter with EAC?

are there differences in quality between fastenc and acm producer pro ? i just saw in encspot that there must be different encoders of fhg. when i encode with steinberg wavlab encspot says that it was encoded with fastenc.

Don't think u can use the filter with EAC.
Of course u can rip with EAC and then compress the wav with Cool Edit.

Yes, on certain samples. For example, try encoding this sample:

http://ff123.net/samples/ma11short.flac

with both encoders. You should hear a clear difference at 192 kbit/s and below (if you leave the ACM encoder in its default joint-stereo setting). However, this may be an anomaly in real music. Most of the time, I'm satisfied with either codec above 128 kbit/s.

At 128 kbit/s, I personally would take fastenc over acm (others would prefer acm). The differences happen often enough at this bitrate that I can have a preference.

ff123

Hmm... I haven't touched MusicMatch in almost five years. If it now includes a full version of the FhG encoder it might be a worthwhile download for certain situations. Does anyone know if there are command-line settings (yes, I am aware that MusicMatch uses its own GUI) so that it could be used with different frontends, or batch files for transcoding? Then maybe folks would finally stop using fastencc.exe....

- M.

Sounds kinda fishy to me. Judging from the mmj download page it mentions the ability to create mp3's from cd vinyl or tape is available in the pro but not basic version.

try this link
[Garf The Nazi moderator was here and recommends you read: http://www.hydrogenaudio.org/forums/index.....=ST&f=1&t=3974]
i think it still works

... better yet, don't. That's an old version of the "Radium" ACM codec. Not desirable.

- M.

So where can we find the latest FhG ACM codec???

I wouldn't mind getting my hands on that...

The latest FhG ACM codec available to the public is the low-bitrate-only codec installed with Windows Media Player... so the odds are you already have it. But if you meant the latest high-bitrate FhG ACM codec, you are out of luck; "not desirable" was a code phrase for "that link is pointing to an illegal/hacked program, and this is not the appropriate place for such material."

- M.

from EAC whatsnew
Posted: Jun, 22 2001 (www.exactaudiocopy.de)


And yes, it works pretty well.

FhG Fastenc is not available as an ACM, and the latest FhG ACM offers no clear-cut advantage over this blazing fast codec (see ff123 tests).

So I'd say that EAC + fht.flt is the best combo for your FhG MP3 Encoding needs.

Oops...

hm fhg codecs are confusing . i think i'll use the fastenc thing included with steinberg wavlab. i want to play around with low bitrates anyway, so...

The last version of Frahunofer is in:
Cool Edit Pro 2.0
Nero Wave Editor
Music Match Jukebox 7*
Steinberg
Magix mp3 maker
Audioactive Production Studio

This latest version always use JS for CBR encoding and Stereo in VBR (only Cool Edit has the option for disable Joint Stereo and lowpass).
In Cool Edit Pro 2.0 and Nero Wave Editor if you set medium or low quality then Fastenc it's used.

Then there are Fastenc (we have 2 version of Fastenc, one has the stereo collapse bug) and Radium ACM Codec (made in 1999).
The demo version available on Fhg site I think it's not the latest.

If i'm not mistaken, "Radium" was the name of the "crack group" that released it, usually in the form of a "Freeware" Producer Pro 2.1 (which was actually commercial). But a non cracked Producer Pro contains this encoder, without the tweaks the Radium group did. Also, i think Divx 3.11a comes with the Radium tweaked ACM encoder (So you could save/load a wav/avi file with an mp3 inside).

Producer Pro 2.1 only comes with the ACM DEcoder.

http://web.onda.com.br/rjamorim/mp3prod.png



That famous fastencc version that can be found in some geocities page <cough> is no more than a stolen library wrapped inside a CLI frontend. More info here:
http://www.hydrogenaudio.org/forums/index....=ST&f=15&t=3670



Right, mp3enc 3.1 doesn't use the latest pro encoding libraries from FhG.

And I would believe Audioactive isn't using them either. They have been offering the same version of AAPS (2.04) for download for ages. :-P

I have Audioactive Production Studio 2 and the output at 128kbps seems to be the same produced by Music Match Jukebox 7 or CEP with default settings ( at least the sound quality is the same).

Have you tried doing a binary comparision of the outputs? That way, we can (hopefully) know if it's the exact same encoder or something else.

I thought it was a 1997 FhG ACM codec, oh maybe it's made in 1999

Maybe you guys should read this decoder test

http://mp3decoders.mp3-tech.org/decoders_a...cm.html#acm1999


EDIT : How you enable stereo in Fastenc????
Only JS in CEP, there's no such stereo switch in CEP 2.0!


do they cost something or can you download them freely ?
MUSICMATCH 7

http://web.onda.com.br/rjamorim/radium.png

Version 4.2 of the FhG acm (released by Krystal Studios) is included in the latest Nimo Codec Pack.

http://nimo.titanesk.com/nimo/
http://sq.onlinedown.net/down/Nimo50Build9Beta1.exe

and there ya go

I've encoded a wav file to mp3 (128kbps) using:
-Cool Edit Pro 2.00 (2095.0) - The settings used in CEP2 are the default settings:
Maximum Bandwidth: 15804 Hz
Codec: High Quality Codec (Slowest)
Set Original Bit
Padding: No Padding
Allow Mid-Side Joint Stereo

-Audioactive Production Studio 2.04
44100 Hz, 'Higher quality encode (Optimize audio quality)'

Music Match Jukebox 7
-Default settings (very high)

The filesize of the wav source is 16.027.652
The filesize of the three mp3 is 1.451.160

1. Comparing the three mp3 with Norton File Compare the mp3 created with Audioactive Production Studio is the longer.

2. Encspot shows that the number of Short Blocks used it's different in all the mp3s.

3.The spectral analysis shows that the mp3 produced by CEP 2, Audioactive and Music Match have a cutoff at 15500 Khz, when for mp3enc, fastenc and radium the cutoff it's different (in Radium for example is near to 19 Khz and the volume is higher than the source).

4.The quality of the produced in the tested programs is the same.

So:
Audioactive Productions Studio Pro 2.04 don't use the same version of Music Match or CEP 2 but sure it's a recent version.

Who knows how to use the Frahunofer fht filter in EAC?

What version of audioactive? I dont know how/why mine would be different (v 2.0.4j), but @128k the default cutoff seems to be 19.5kHz - 20kHz.

Mine also has the same cutoff at 160k with a tad less highfrequency above 16kHz being masked away, and then at 192k it goes full spectrum 22kHz.

Is this not the case? Hmm... Im going to do some testing with that program you used to check out the mp3, see what it tells me, take some screenies, and get back with results if different.

j

Whish one of these FhG encoders is the latest - they are both included in the Nimo Codec Pack:

Fraunhofer IIS MPEG Layer-3 Codec (professional) - This codec is labeled as v1.2 - Build 63i

OR

MPEG Layer-3 Codec (professional) - This codec is labeled as v4.2 and is copyrighted by Kristal Studios (If i'm correct, this is also refered to as the Radium codec v1.2.63)

Are they the same codec? I set them both to simple stereo and encoded the sweep.wav file @ 192kbps at both high and low quality in BOTH codecs. The high quality modes of both codecs produce the exact same file, but the low quality (fast) modes give two different files (they are not even close - different Av. / Max Reservoir levels and different scalefac values) Which one of these codecs is of higher quality, and at what mode (low / high quality setting.)

thats FhG 1.263, radium version


thats FhG 4.2


compare 1.2 and 4.2 at 128k or below.... remember that v4.2 has a cutoff of ~20kHz at 128k. while radium is at 16kHz. at 160k they seem to be about the same on HQ, below 160k v4.2 seems to handle jointstereo better and has the higher cutoffs - bonus. B)

Hold up.

Isn't the latest Lame ACM already better than the newest ACM by FhG?

Anybody care to respond?

3.94a11?
or 3.90.2 dibroms, heheh see... its hard to say what's the "best" lame codec just as it is hard to say what is the best fraunhoffer, because there will always be slight differences that some people precieve more easily than others.

in my opinion, Audioactive 2.0.4j is the reference FhG encoder to use. compare CoolEditPro2 to it at 160k and see what I mean. But my runner up is definately CoolEditPro2 (decent configurability)

I think lame394a11 definately competes with Aa204j. Ive heard it beat Aa on several samples designed to reveal pre-echo -- but -- the number of frequencies across the range always seems fuller with FhG, despite the smaller filesize. (imo)

To my knowledge, the quality of MMJB's MP3 encoder at the "High" and "Very High" quality settings is horribly flawed. It mangles all parts of the sound - just take a file such as sweep.wav and encode it at 160 - 256kbps. You can see the dropouts and hear how badly it is flawed. Try to ABX it, it should not be hard at all. The normal setting seems to have the highest quality of the three. This was discussed at one point on r3mix.net, and even though that site is looked down upon now, their findings on this subject still seem to hold true. I tested using the sweep.wav file, and it is easy to see the sound corruption (tested using MMJB v7.20.0173 - I don't know if the 7.5 versions still have the problem.) Also, 128 / 320kbps modes don't seem to have the problem (128kbps cuts out the mangled frequencies / 320kbps seems to default to "normal" mode for some reason ) Could someone possibly verify this problem?

http://www.geocities.com/syntax_ws2/
installer for just FhG version 4.2 mentioned above as part of the Nimo pack.

As said at the start of the post, the version is 2.04

And for Music Match Jukebox I've encoded some samples for the test I made and the sound isn't too bad and seems to be the same of CEP 2 and Audioactive Production Studio (I've not tried killer samples though).


Lame ACM it's better and I suppose it use alt-preset.

AudioPro12,
Thanks for the file, someone was just asking about mp3 codec today, I hate to make them download Nimo just for it. B)

[edit]
Oge_user,
the reason I specified 2.0.4j was because theres nearly a year between the original 2.0.4 and the j revision. I have absolutely no idea what the actual differences are although.
[/edit]

If the latest Lame ACM was made in 2000 or 2001, then it's not too great. If there's actually been a new one made this year or last, it should really be good.

lame 3.94 alpha 11 actually...
far as i know this is the most current acm (feb 28 2003) and often updates daily
http://homepage.ntlworld.com/jfe1205/lame3.94a11.zip

ViPER1313:

I thought that, from MMJB 6.1 onwards, both FastEnc and Alternate codecs were finally "clean" (in MMJB 6.1 they certainly both were OK). I would be very surprised if any of those severe bugs re-appeared in a later version.

FYI, the problems I remember with the faulty Alternate codec ("high" / "very high" setting), as found in MusicMatch 4.x and 5.x (5.1.1052 IIRC), were mainly really extreme ringing. Try encoding some real music and listening to it.

(I'd download the app and check for myself if I had a fast connection, but... )

[EDIT] Just in case Wish is reading this thread - I swear that 6.1 was the last version of MMJB I have ever touched, and that's more than a year ago. [/EDIT]

I've made a test using the Frequency Analysis in Cool Edit Pro, so we can at least identify which versions are similar (newer versions use a lowpass of 15,5 khz).

The bitrate is 128kbps with the default settings (Audioactive use by default 48000 Hz but I have set it to 44100 Hz). The source file is bird.wav

Radium 1.263 (1.2 Build 63i, Copyright Frahunofer-Gesellschaft IIS)


Fastenc 1.01 (1994-2000) (Commandline version, highest quality


Frahunofer 4.2 (Copyright Krystal Studio 1999-2000)

Note: Like already said, this version is also contained in Nimo Pack 5.

Audioactive Production Studio 2.04


Music Match Jukebox 7.5


Cool Edit Pro 2.00


In my opinion the best quality is given by the 4.2 version.
Even if with Cool Edit you can disable the internal lowpass of Frahunofer (just set it to 22050) and apply the Chebychev Lowpass (set it at 16,5 Khz) obtaining a good quality better than 4.2.

I've not tested the 3.1 (it's available on FHG site).

Its interesting how radium is the only version that encoded above 16kHz. I ran some balanced white noise through Audioactive 204j and came up with this:

Why didnt it encode bird.wav properly?

AFAIK they are have 2 type of ACM Codec




l3codeca.acm - "advanced"
decodes all bitrates. Only encodes up to 56kbps. (Windows ship with this one)

l3codecp.acm - "professional"
encodes and decodes all bitrates

And they have 2 version , 1.0.xx(1997-buggy) and 1.9.x.x (1999-newest)


Version I am using is


and I believe this is newest from Fraunhofer



4.2 xxx version from Nimo is a different story

BTW it's come from Media Juke Box

1, 9, 0, 0305
11 - 22 - 2000

4, 2, x, x x
3 - 2 - 2001


not a bit difference in time, so the version number seems kinda whack... what's the story morning glory??

Yes, also Audioactive encode frequencies above 16Khz:
-I've encoded other tracks with Audioactive and the lowpass is of 20Khz.
-Same problem with Frahunofer Krystal version: only for the 'birds' sample the lowpass is 15 Khz, with other tracks is of 20Khz.
-With the other version no problems, the lowpass showed is the effective lowpass of the programs.

Seems that Audioactive and Krystal made the same mistake on the same track, maybe they use the same version, what do you think?

The sample I used I've downloaded from the topic Test sample archives, try to encode it with Audioactive to see if your version has the same problem.

Hello
Oge

winamp reports incorrect track lengths for tracks vbr encoded with the FhG filter from cool edit pro2
anyone know how to fix that?

If you used VBR, VBR Fix (http://www.willwap.co.uk/) will resolve the problem.

oops..delete me plz

Need some clarity here, from some folks who've got a little more experience with fastencc than I do, and who know all its ins-outs-quirks. (FWIW, I already searched the forum and couldn't find a suitable - or even consistent - answer to this.)

fastencc 1.02 is reported to have a stereo collapse bug in the fast mode, but what about the -hq mode? So far the only "bug" I've seen described in association with that mode is the slow speed... which would be expected if the libraries being used for that mode were more along the lines of FhG's MP3Enc (i.e., the "professional" libraries).

So out of curiosity I hunted up a copy and ran a few quick tests. Speed-wise here's what I found (the test system was a PIIIM 1.06GHz notebook):

At -br 128000 -hq, for anything that had a wide stereo separation the speed hovered around 1.1x.

At -br 128000 -hq, for anything that had a very narrow stereo separation the speed hovered around 8.1x.

For a very low-bitrate mono encode (such as would be used for speech) at -br 32000 -ds 22050 -dm -hq, the speed soared to 47.9x for the wide-stereo sources, and 48.1x for the sources which were already close to monophonic. (Same sources as above.)

So here is the question: For material which can still be appreciated as a low-bitrate, monophonic stream, and when a CBR stream is required, would there be any compelling quality-based argument against using fastencc in -hq mode?

- M.

Edit: Fixed a few stupid spelling errors.

sure M -hq will always encode in better quality of that version of fast. btw, using fastenc with -hq is just envoking the normal fast codec, instead of the fast-fast codec. get it?

to test the latest available fast codec, use CoolEditPro2.1 fastenc. Ive already heard one FhG person mention that it's thier favorite codec/batch-program for thier own use... and FhG improved speed and quality considerably since they last released the free commandline version.

almost forgot about this thread

j


[edit] if you test the new codec, remember that the p3 isnt going to use SSE2 and quite a few other optimizations of pentium4/athlonxp or later [/edit]