Feature request: faac as an additional encoder

Discuss the current and future development of Max.
Post Reply
Julian
Posts: 7
Joined: Tue Apr 10, 2007 2:56 pm

Feature request: faac as an additional encoder

Post by Julian » Tue May 22, 2007 1:43 pm

I like Max alot, i use it for all my music conversions... except when i want to make some high-end purely VBR aacs.

Could faac be added to Max, with settings like cutoff frequency and quality quantizer enabled? It would make Max the perfect program for me.

faac is available to mac via Darwinports.

RonaldPR
Posts: 433
Joined: Tue May 30, 2006 8:27 am
Location: Amsterdam, Netherlands

Re: Feature request: faac as an additional encoder

Post by RonaldPR » Tue May 22, 2007 2:17 pm

Julian wrote:I like Max alot, i use it for all my music conversions... except when i want to make some high-end purely VBR aacs.
Is none of the available MPEG4-audio AAC settings adequate? I thought that FAAC was just a free AAC encoder, not necessarily better than AAC encoding by Core Audio.

Julian
Posts: 7
Joined: Tue Apr 10, 2007 2:56 pm

Post by Julian » Tue May 22, 2007 2:22 pm

Sorry, i didn't mean to say that CoreAudio is bad, in fact i believe it's pretty good, but i don't exactly know what it's doing. It produces a 192 kbps aac at "high quality"... so what? Is that about the bandwith or the (low) number/severity of artefacts?
When using faac, i can just tell it to make the bandwith 0 to 20050 Hz, and reduce the artefacts strongly with quality quantizer 250. When using CoreAudio in Max, i actually have no clue about what i'm producing.

RonaldPR
Posts: 433
Joined: Tue May 30, 2006 8:27 am
Location: Amsterdam, Netherlands

Post by RonaldPR » Tue May 22, 2007 3:19 pm

Julian wrote:Sorry, i didn't mean to say that CoreAudio is bad, in fact i believe it's pretty good, but i don't exactly know what it's doing. It produces a 192 kbps aac at "high quality"... so what? Is that about the bandwith or the (low) number/severity of artefacts?
I am no expert as you appear to be, but in Max I see settings up to 256 kbps VBR maximum quality MPEG4 AAC.

Julian
Posts: 7
Joined: Tue Apr 10, 2007 2:56 pm

Post by Julian » Tue May 22, 2007 3:43 pm

It's not just about quality, it's about a high quality against an as low possible bitrate, in the way that i like.

If you choose 256 vbr with coreaudio in max you get files with bitrates in between 250 and 262 while you don't even know for sure that the bandwith includes 19+ kHz frequencies, if you encode them with faac with settings -q 250 -c 20050 you get files ranging from 170 kbps to 250 kbps and those are all good enough for me.

Maurits
Posts: 117
Joined: Sun Jan 29, 2006 1:36 pm
Location: London, Europe

Post by Maurits » Wed May 30, 2007 9:43 am

FAAC may give you extra tuning options, I very much doubt they will offset the fact that FAAC is just not such a good encoder.

FAAC was tested against Apple's and Nero's AAC encoder over three years ago on HydrogenAudio. Even then, it was no match for both. Worse still, development on FAAC has been pretty much non-existent since (its main contributor has been working for Nero for years now) while both Apple and Nero continued to enhance their encoder.

FAAC just isn't in the same league (anymore).

Any Quality difference between NERO AAC and FAAC?

Julian
Posts: 7
Joined: Tue Apr 10, 2007 2:56 pm

Post by Julian » Wed May 30, 2007 12:28 pm

AAC Encoders - Hydrogenaudio Knowledgebase wrote:FAAC is a free LC AAC encoder under the Lesser GPL license. Its quality has improved drastically over the last few years and FAAC is nowadays a viable alternative to the commercial encoders (although, at 128kbps or lower bitrates, not at the same quality level as some of them, according to Guruboolez's last listening test).
Read it here.

So, faac is quite as good as CoreAudio/Nero at bitrates above 128, which some people surely want to use (me, for example). And as you said, it provides fine tuning possibilities that CoreAudio doesn't currently have in Max. But most important to me, it provides true VBR; the bitrate entirely depends on a fixed quality setting.

Mike1
Posts: 201
Joined: Thu Sep 21, 2006 3:55 pm

Post by Mike1 » Wed May 30, 2007 3:28 pm

Julian wrote:
AAC Encoders - Hydrogenaudio Knowledgebase wrote:FAAC is a free LC AAC encoder under the Lesser GPL license. Its quality has improved drastically over the last few years and FAAC is nowadays a viable alternative to the commercial encoders (although, at 128kbps or lower bitrates, not at the same quality level as some of them, according to Guruboolez's last listening test).
Read it here.

So, faac is quite as good as CoreAudio/Nero at bitrates above 128
That "So" is making a somewhat large jump between an interpretation of some (very old) test results and a rather large claim: "quite as good as ..."

The encoder may be usable at higher bitrates for at least some samples. (We already knew that from the thread Maurits referenced. But that is not equivalent to being "quite as good as" Core Audio or Nero's encoder at all. And there is no way such a claim could be based on those test results, anyway, since they don't test difficult samples at higher bitrates and, besides, are old.

Julian
Posts: 7
Joined: Tue Apr 10, 2007 2:56 pm

Post by Julian » Wed May 30, 2007 3:42 pm

The section on faac on the page that i referred to was last updated in 2006 (it's actually just as old as the thread Maurits referred to), not that old in my humble opinion.

But suppose faac would be a really bad encoder. Shouldn't users of Max still have the choice whether to encode (semi-) CBR in proprietary CoreAudio or true VBR in free faac?
CoreAudio is probably slightly better in mp3 than LAME as well, so why does Max only provide LAME mp3 and not CoreAudio mp3? Max should either only use CoreAudio, or only free encoders such as faac, LAME and flac, or, what i prefer, both. Not free encoders for some codecs and CoreAudio for others.

Maurits
Posts: 117
Joined: Sun Jan 29, 2006 1:36 pm
Location: London, Europe

Post by Maurits » Wed May 30, 2007 6:09 pm

Julian wrote:The section on faac on the page that i referred to was last updated in 2006 (it's actually just as old as the thread Maurits referred to), not that old in my humble opinion.
Yep, and it's either a very old quote or they have an odd notion of time. The last serious quality update on FAAC was in February 2004. Even then it was beaten hands down by both Apple's and Nero's encoders. And they have both seen a lot of further improvements since. FAAC's development has stalled long ago where the other AAC encoders have not. Even FAAC's own website seems to have been taken down.
But suppose faac would be a really bad encoder. Shouldn't users of Max still have the choice whether to encode (semi-) CBR in proprietary CoreAudio or true VBR in free faac?
If there was a possibility of getting a better VBR implementation that would be great. I just don't see why FAAC would deliver that. You choose a (good) VBR algorithm because you opt for quality. Why would you choose a known worse encoder because it happens its VBR implementation works theoretically better? That VBR algorithm doesn't seem to help against Apple and Nero when only quality is a concern, in a blind listening test. Granted, if Nero's encoder could work on OS X it would be worth considering as it has certain benefits over Apple's, including a more variable VBR implementation. Sadly, it is not an option right now.
CoreAudio is probably slightly better in mp3 than LAME as well, so why does Max only provide LAME mp3 and not CoreAudio mp3? Max should either only use CoreAudio, or only free encoders such as faac, LAME and flac, or, what i prefer, both. Not free encoders for some codecs and CoreAudio for others.
CoreAudio's MP3 encoder is known to be not that good. Allegedly it has gotten better lately but there is no evidence it comes close to LAME. There is no reason to have it as an option.

The encoders Max uses are there because of their quality, they are known (after much serious testing!) to be the best for that particular format, they are not chosen for political reasons.

User avatar
sbooth
Site Admin
Posts: 2445
Joined: Fri Dec 23, 2005 7:45 am
Location: USA
Contact:

Post by sbooth » Wed May 30, 2007 6:18 pm

Julian wrote:CoreAudio is probably slightly better in mp3 than LAME as well, so why does Max only provide LAME mp3 and not CoreAudio mp3? Max should either only use CoreAudio, or only free encoders such as faac, LAME and flac, or, what i prefer, both. Not free encoders for some codecs and CoreAudio for others.
Actually, there is no Core Audio MP3 encoder. Core Audio provides an MP3 decoder (which honestly isn't that great which is I why I use MAD in Play). iTunes uses its own MP3 encoder.

From the Core Audio documention:
Core Audio generally expects audio data to be in native-endian 32-bit floating-point linear PCM format. However, you can create audio converters to translate audio data between different linear PCM variants. You also use these converters to translate between linear PCM and compressed audio formats such as MP3 and Apple Lossless. Core Audio supplies codecs to translate most common digital audio formats (though it does not supply an encoder for converting to MP3).

User avatar
sbooth
Site Admin
Posts: 2445
Joined: Fri Dec 23, 2005 7:45 am
Location: USA
Contact:

Post by sbooth » Wed May 30, 2007 6:28 pm

Julian wrote:But suppose faac would be a really bad encoder. Shouldn't users of Max still have the choice whether to encode (semi-) CBR in proprietary CoreAudio or true VBR in free faac?
I'm not really knowledgeable as to whether faac is as good as, worse than, or better than the Core Audio AAC encoder. So I will abstain from that discussion.

To address your point above philosphically, my goal with Max is to provide the highest-quality audio that is possible for a given format. It is why I chose aoyumi's tuned version of vorbis, and LAME for MP3. So if faac truly was really bad (which I am not asserting), I would not include it with Max. I would not do so because the average user doesn't understand audio compression or encoders- they would simply say "Max creates AACs that sound like crap".

That being said, if faac was comparable to Core Audio, I would not be averse to including it in Max if the license was GPL-compatible and I found the time to add it.

Mike1
Posts: 201
Joined: Thu Sep 21, 2006 3:55 pm

Post by Mike1 » Wed May 30, 2007 9:51 pm

sbooth wrote:That being said, if faac was comparable to Core Audio, I would not be averse to including it in Max if the license was GPL-compatible and I found the time to add it.
The codecs are released under different licenses: while FAAC and FAAD (the outdated first release of the decoder) use the LGPL, FAAD2 is GPL. For more information see the homepage and/or the Readme files in the source code package or CVS, also about patent issues related to the AAC format.
http://freshmeat.net/projects/faac/

It's difficult to know what's going on with FAAC - if anything. Audiocoding is no longer accessible:

http://www.audiocoding.com/

The software's still available, but I have seen a few complaints recently about people having difficulty getting it to work properly, or at all, on Linux (specifically Ubuntu/Kubuntu 7.04, which is what everyone seems to use these days). There seems to be an issue with it making screeching noises and skipping with Sound-juicer, but maybe that's down to Sound-juicer.

I have got it going on a nix box myself. At the recommended quality setting it encoded an Eric Clapton track at 181kbps, whereas using Max Core Audio 160kbps VBR gave it 168kbps. Comparable filesizes/bitrates, IOW. I thought the Core Audio file sounded a little better, but I didn't ABX them, and haven't got the software to do that.

Julian
Posts: 7
Joined: Tue Apr 10, 2007 2:56 pm

Post by Julian » Wed May 30, 2007 10:23 pm

Mike1 wrote:The software's still available, but I have seen a few complaints recently about people having difficulty getting it to work properly, or at all, on Linux (specifically Ubuntu/Kubuntu 7.04, which is what everyone seems to use these days). There seems to be an issue with it making screeching noises and skipping with Sound-juicer, but maybe that's down to Sound-juicer.

I have got it going on a nix box myself. At the recommended quality setting it encoded an Eric Clapton track at 181kbps, whereas using Max Core Audio 160kbps VBR gave it 168kbps. Comparable filesizes/bitrates, IOW. I thought the Core Audio file sounded a little better, but I didn't ABX them, and haven't got the software to do that.
I had some problems playing aacs from faac too in QuickTime. They appeared to be due to QuickTime, as there was no problem in iTunes.
Of course, this still doesn't tell much about the quality of faac. I'm now very eager to find out for myself (compare CoreAudio and faac), so i'll probably download some software for double-blind testing later this week.

In any case, i have been using faac with great pleasure uptill now and i have the feeling that my new faac -q 250 -c 20050 files sound alot better than my old 192 kbps iTunes aac tracks. When i have more objective results i'll post them here.

Post Reply

Who is online

Users browsing this forum: No registered users and 2 guests