Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: FAAC 1.24 released (Read 7990 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

FAAC 1.24 released

FAAC 1.24 has been released. See the audiocoding.com Wiki for details on how to use the new interface.

FAAC 1.24 released

Reply #1 –
The new features are:

Code: [Select]
Frontend changes:
- MP4 metadata support
- ABR now uses total bitrate, and not bit rate per channel
- command line interface completely revamped, see built-in documentation
for details
- Temporal noise shaping, TNS, no longer disabled by default

Library changes:
- default low pass values for quality-based VBR mode


It's worth noting that the Wiki is slightly out of date - but Hans-Jürgen said this would be fixed shortly. Also, all frontends will need to be updated to the new interface - it's completely incompatible with the old one.

By the way, if any of you have huge knowledge about mis/side coding and/or your own psychoacoustic model, ready for integration in FAAC, then you should contribute to the project  It seems that FAAC failed mainly on the mid/side intensitive samples in Roberto's test, but did well on the rest.

I only wish I had the ears to actually tune it

FAAC 1.24 released

Reply #2 –
The win32 binaries at Rarewares are already updated (as of yesterday!!  ).

FAAC 1.24 released

Reply #3 –
Quote
The win32 binaries at Rarewares are already updated [...]


"with our powers combined!"...

actually, teh RW Debian Rep is updated since the last beta packaging, since the only change since was the version re-brand (but upped anyways, for teh picky)


later

FAAC 1.24 released

Reply #4 –
great news that you work on faac!
i hope some quality tunings will follow too

now that faac supports mp4 and metadata (i assume this means itunes tags) may i bring up a similar issue:
as you propably know mpeg-7 now specified how to place tags in .mp4, and yes its not the way itunes uses
is it possible to get support for official tagging in faac? maybe a switch which lets someone choose which tagging format (or both) is wanted would be great!
I know, that I know nothing (Socrates)

FAAC 1.24 released

Reply #5 –
Quote
great news that you work on faac!
i hope some quality tunings will follow too

I would very much like to tune FAAC. However, there is one problem: I can't. Other than the fact that I don't know much about the core of FAAC, I couldn't ABX velvet @ q100 (best score: 2/2) which most testers said sounded horribly. So I try to help in the ways I can. It's worth noting that the VBR low pass values in FAAC are completely untuned - if someone believes some other values would be better, please say so.

Quote
now that faac supports mp4 and metadata (i assume this means itunes tags) may i bring up a similar issue:
as you propably know mpeg-7 now specified how to place tags in .mp4, and yes its not the way itunes uses
is it possible to get support for official tagging in faac? maybe a switch which lets someone choose which tagging format (or both) is wanted would be great!

The MPEG-4 metadata code in FAAC is nothing but calls to libmp4v2, and it doesn't have support for MPEG-7 yet. Are there any MPL/LGPL-compatibly licensed libraries out there that implement MPEG-7 metadata? Speaking of which, do any players support it?

FAAC 1.24 released

Reply #6 –
MPEG-7?  Never heard of it.  iPod compatibility is orders of magnitude more important IMO.

Lowpass:  The  c values are high compared to what other codecs are using.  You may be wasting a few bits.  However, it might do a bit better in a 128k shootout because of the higher default c value.  In Roberto's test there was no attempt to tune the c value so the default of 16k ws used.

Psy models:  how about looking over the mpc source code, now that it is an open source project.  At least you know the problem is in the mid side area, that is halfway to a solution.

I encoded  few things with john33's latest build.  It works.  Speed wise it is about midway between non P4 optimized versions of vorbis and lame 3.96.

FAAC 1.24 released

Reply #7 –
Quote
Quote
now that faac supports mp4 and metadata (i assume this means itunes tags) may i bring up a similar issue:
as you propably know mpeg-7 now specified how to place tags in .mp4, and yes its not the way itunes uses
is it possible to get support for official tagging in faac? maybe a switch which lets someone choose which tagging format (or both) is wanted would be great!

The MPEG-4 metadata code in FAAC is nothing but calls to libmp4v2, and it doesn't have support for MPEG-7 yet. Are there any MPL/LGPL-compatibly licensed libraries out there that implement MPEG-7 metadata? Speaking of which, do any players support it?

hm i dont think that there are any libraries out there supporting mpeg-7 tags already
afaik this was standardized just a short time ago (dont even know if it is finished already)

Quote
MPEG-7?  Never heard of it.  iPod compatibility is orders of magnitude more important IMO.

well support for the official tagging standard doesnt mean that the support for apples itunes tags will get removed 

also from what i saw the itunes tags are very limited and very audio-centric (yeah video is possible in mp4 too  ), also i didnt see a tool till now which allows to itunes-tag two audio streams in one mp4 file seperately for example (is this possible?)
well from what i saw till now the itunes tags can maybe be seen as yet another example of the cases which show that things which arent thought trough and standardized by mpeg are very often very limited (i hope someone can proove me wrong on this)
I know, that I know nothing (Socrates)

FAAC 1.24 released

Reply #8 –
i second the move to take a look at mpc source

i may be talking rot, but wasn't there some talk about mpc's psy-model becoming a separate thing, kinda like a "pluggable" psy model?  if this is the case then FAAC could benefit a lot from this (what i understand about mpc's psy model, it could be adapted to AAC's limits, but would need some hacking at to get the sfb's right)

FAAC 1.24 released

Reply #9 –
Quote
also from what i saw the itunes tags are very limited and very audio-centric (yeah video is possible in mp4 too  ), also i didnt see a tool till now which allows to itunes-tag two audio streams in one mp4 file seperately for example (is this possible?)
well from what i saw till now the itunes tags can maybe be seen as yet another example of the cases which show that things which arent thought trough and standardized by mpeg are very often very limited (i hope someone can proove me wrong on this)

IMHO the iTunes tags are intended for a different purpose than MPEG-7. MPEG-7 is intended to be a information layer to supplement the audio and video, whereas the iTunes tags are like Ogg and APE tags: File specific, easily parsed and easily extensible. AFACT a MP4 file is pretty much percieved the same way in FAAC as iTunes: One file, one song, one set of metadata.

Quote
i second the move to take a look at mpc source

So do I. Unfortunately, it seems to require much more knowledge of audio coding than I possess. If someone else were willing to take a shot, I'd help in every possible way I could

FAAC 1.24 released

Reply #10 –
About the lowpass filtering: I'd say it would be wise to limit the auto adjustment to 20 kHz even at very high q-values. If someone feels like needing more... -c 

btw: Does faac use a smooth fading (called poly-phase lowpass in Lame)? Useful?

FAAC 1.24 released

Reply #11 –
I know that FAAC is not the best aac encoder around,
but is the quality comparable to LAME?
Wanna buy a monkey?

FAAC 1.24 released

Reply #12 –
Quote
I know that FAAC is not the best aac encoder around,
but is the quality comparable to LAME?

I'll let a Lame developer answer:

http://www.hydrogenaudio.org/forums/index....ndpost&p=190466

Quote
Lame is full of tricks everywhere, and a simple aac codec is performing better!


Of course, that comment is based on my listening test results, so it only really applies to 128kbps. It wouldn't be safe to extrapolate it to higher bitrates without some good testing.

FAAC 1.24 released

Reply #13 –
Are there plans to add presets to FAAC? Something like:

streaming
transparent
audiophile
flac > schiit modi > schiit magni > hd650

FAAC 1.24 released

Reply #14 –
Quote
About the lowpass filtering: I'd say it would be wise to limit the auto adjustment to 20 kHz even at very high q-values. If someone feels like needing more... -c  

btw: Does faac use a smooth fading (called poly-phase lowpass in Lame)? Useful?

polyphase is just the algorithm...nothing to do with the smoothness of the filter, iirc it's quite steep in lame just like in other codecs
 

FAAC 1.24 released

Reply #15 –
Why the "UNSTABLE" tag? Has that always been there and I just missed it?

bitrate = 204
samplerate = 44100
channels = 2
aac_profile = LC AAC
codec = AAC
tool = FAAC 1.24+ (May  7 2004) UNSTABLE
replaygain_track_gain = -6.82 dB
replaygain_track_peak = 1.118449
replaygain_album_gain = -6.74 dB
replaygain_album_peak = 1.132592
----------
13809180 samples @ 44100Hz
File size: 8 060 774 bytes
flac > schiit modi > schiit magni > hd650

FAAC 1.24 released

Reply #16 –
UNSTABLE = You are using a version compiled from cvs, which is not considered released, thus it's marked as unstable. Of course this does not mean that it's bad, just that it might be bad.

(Just a guess by looking at the compilation date)
Life is Real...
(But not in audio :) )

FAAC 1.24 released

Reply #17 –
It's the version from Rarewares.
flac > schiit modi > schiit magni > hd650

FAAC 1.24 released

Reply #18 –
Quote
It's the version from Rarewares.

If you want a version without the unstable tag, you need to compile the sources at audiocoding.com.

FAAC 1.24 released

Reply #19 –
well, so far dan christiansen has been adding/fixing non-quality based changes to cvs

...so it being "UNSTABLE" just means it isn't a release, and should be fine for faac atm



later

FAAC 1.24 released

Reply #20 –
Quote
Quote
It's the version from Rarewares.

If you want a version without the unstable tag, you need to compile the sources at audiocoding.com.

It doesn't bother me, I just wanted to make sure I was using the right version. Not being a coder, I don't even know what I would need to compile it. I've compiled applications in Linux, but never on a Windows platform. On Linux it was always just

./configure
make
make install

I'd like to learn how to compile some of these encoders on Windows. Any good refernces?
flac > schiit modi > schiit magni > hd650

FAAC 1.24 released

Reply #21 –
Quote
On Linux it was always just

./configure
make
make install

It's exactly the same for faac in windows if you use cygwin or mingw.

FAAC 1.24 released

Reply #22 –
You can check at this link, some guidelines for compiling under MS VC6. I think there are already proper MS VC7 project files included with the source download, never tried them though.

And, I also tried mingw compilation, but it gave me half the speed compared to my VC6 compile. And I could not compile faad2 and libmp4v2, so no mp4 encoding for me.
Life is Real...
(But not in audio :) )