PDA

View Full Version : Fresh build of reference Enhanced AACPlus (AACPlus v2) Codec


dimzon
22nd August 2005, 18:57
New compile (http://s18.yousendit.com/d.aspx?id=0BD951NLBULCM3T4VCZKJZ2110) (VS.NET 2003) (fixed point encoder and floating point encoder)

Previous version (http://forum.dbpoweramp.com/showthread.php?t=7107)

Sources (http://www.3gpp.org/ftp/Specs/html-info/26410.htm)

buzzqw
22nd August 2005, 22:08
any link to read more info about this encoder ?

BHH

SeeMoreDigital
22nd August 2005, 22:32
Does this codec only work with dbpowerAMP?

I would be interested to know if a GUI could be developed... We could do with more encoders offering AAC-HE... And one that could offer a 2-pass option!


Cheers

Hyper Shinchan
23rd August 2005, 01:44
Hi guys! It's a long time since my last post.....
anyway, I agree with Buzzqw, I've discovered this encoder yesterday on the FAAC website, and it's an interesting encoder....
If i've well understand this is a compiled version of the AAC plus v2 demo of the 3GPP developpers.
But why there are 2 encoders (one called enhAacPlusEncFX and the other enhAacPlusEncFP)? 2 different version? What are the differences?
Thank you in advance.
P.S.: but this encoder is CBR or ABR? And to decode AAC plus v2 I can use the same decoder that I use for AAC-HE (or AAC Plus), right?

dimzon
23rd August 2005, 10:23
buzzqw
You can read some discussion here (http://forum.dbpoweramp.com/showthread.php?t=7107), here (http://www.hydrogenaudio.org/forums/index.php?showtopic=33645) and here (http://www.hydrogenaudio.org/forums/index.php?showtopic=34560).

SeeMoreDigital
It's command line tool so You can use it from command line. In previous (http://forum.dbpoweramp.com/showthread.php?t=7107) build it was bundled with wrapper for dbPowerAMP GUI.

Hyper Shinchan
enhAacPlusEncFX is fixed point implementation (faster)
enhAacPlusEncFP is floating point implementation (slower but, maybe, newer)

Hyper Shinchan
23rd August 2005, 11:42
Thank you dimzon, from my first tests is really an interesting encoder, with very low bitrate (24kbkbps stereo and lower) the audio quality is really better than the Nero AAC-HE encoder.
P.S.: in the next realease can you include something like a readme file, for incompetent users like me ;-) ?

CiNcH
23rd August 2005, 17:18
This compile does not support 44.1 KHz input which has been added by the dBPowerAmp developer to his version of the encoder.

Oki
23rd August 2005, 18:38
Thank you dimzon, from my first tests is really an interesting encoder, with very low bitrate (24kbkbps stereo and lower) the audio quality is really better than the Nero AAC-HE encoder.
P.S.: in the next realease can you include something like a readme file, for incompetent users like me ;-) ?This encoder uses HE-AAC v2, it means that it uses the MPEG Parametric Stereo tool for low bitrates under 36Kbps, giving a lot more quality than an HE-AAC encoder without that tool.

Regards,
Oki

DeathTheSheep
23rd August 2005, 19:05
CiNcH: Yes, 44.1khz is all but a requirement these days...what audio (music files, etc) besides DVD content is encoded in 48khz?

But anyways, yes, I guess a fresh compile was in order. Thank ye much. ;)

However, it appears as if the Fixed Point encoder is broken-- it is unplayable by Nero or faad2lib and it shows this message after encoding:
(file only has 5318 samples!!)

[5319]
end of TheOneFX.3gp reached


encoding finished
The floating point encoder displays a different (shorter) intro message, takes much less time to encode, and produces a working file. The encoding process displays this:

[5318]

encoding finished
Perhaps it's the fault of the decoder(s), but I think that the FX encoder is bonkers and goes overboard when encoding.

I'm really anxious to test it out, though, for it seems newer and a bit slower (so, even though fixed point calculation is naturally slower, perhaps it yeilds significantly better quality). But, it simply doesn't work with any input I give it (48khz, 2-channel 16-bit PCM audio)--it overflows and produces output that cannot be decoded.

Mug Funky
24th August 2005, 11:25
Yes, 44.1khz is all but a requirement these days...what audio (music files, etc) besides DVD content is encoded in 48khz?

almost everything but CDs... SACDs are the odd one out - IMHO an unworkable format. 48k gives just enough extra above the audible range, but doesn't sacrifice accuracy (there's no such thing as true 96khz, 24-bit. as rate increases, precision suffers. the point of diminishing returns for 24-bit is around 55Khz or so IIRC).

DVD was chosen as 48k because it was already standard - all the sources for DVD are also 48k - Digibeta, and all flavours of DV (except 4ch DV which is 32k, 12-bit in the same space as 48k, 16-bit). archive services also request master be 48k, 24-bit if possible.

btw, this encoder looks real interesting. must try it out at some point. it'd be kickass to have it tuned as tightly as LAME some day.

DeathTheSheep
24th August 2005, 21:42
That's pretty interesting. I had no idea the 48khz range was so ... standard.

I'd thought that since it seems that over 90% of compressed songs (illegal or not) are CD rips, and all have 44.1k rate w/16bit precision.

But okay, I guess if 48k is the standard, it shouldn't be so hard to convert a 44.1k song into a 48k one. You'll just loose a bit of quality (unless you have a really good anti-aliasing/artifacting filter).

Cheers

dimzon
25th August 2005, 11:08
I think AAC+ 48000 is better than AAC+ 44100 (because AAC+ nature)

DeathTheSheep
25th August 2005, 22:33
Cool. ^_^
Any reason why the FX encoder doesn't stop after it reaches the last sample in the file? Or why the resulting output doesn't play back with Nero or FAAD2? I really want to test FX's quality, but I keep running into this strange problem :eek:!

Hyper Shinchan
25th August 2005, 23:46
This encoder uses HE-AAC v2, it means that it uses the MPEG Parametric Stereo tool for low bitrates under 36Kbps, giving a lot more quality than an HE-AAC encoder without that tool.

Regards,
Oki
I know this ^_^. I was talking about the usage (e.g.:the 48 khz input only, I've touse the db build, it's not a good idea to add the 44.1 khz input? At the moment I prefer to use the old build).
And about the fp and fx points difference.
But I want to know more about this: when you open the file encoded in AAC Plus v2 with MP4UI, it says that it's MPEG-4 scalable@L1, but when you open normal AAC-HE (without the PS) it says that it's MPEG-4 Main@L2 (AAC LC) (probably it doesn't recognize the HE for backward compatibility, I think).
But MPEG-4 scalable it's the same than AAC+SBR+PS? I've read something about it in the Fraunhofer website, but they talk about them in 2 different pdf.
What is the difference (if there is a difference)?

Rockaria
26th August 2005, 06:41
As for identifying the profile, inspecting the demuxed AAC(with MP4creator) in the foobar2K was the closest one so far(still no PS) to me.
Being not so much a mobile user, I am more interested in the AAC-LC transcoding to 48k 900~kbs files because of the system mixer nature and the spdif-out nature.

Btw, anybody know what is the advantage of the capture device enforcing the spdif-in to be 44.1k pcm?

ggab
26th August 2005, 08:43
yes, the 44100hz wav input fails:

C:\Mis documentos>enhAacPlusEncFP.exe 001.wav 123.mpa 48000 s

*************************************************************
* Enhanced aacPlus 3GPP Floating Point Reference Encoder
* Build Aug 22 2005, 17:01:58
*
*************************************************************

No valid SBR configuration found


C:\Mis documentos>enhAacPlusEncFx.exe 001.wav 123.mpa 48000 s

*************************************************************
* Coding Technologies' enhanced aacPlus 3GPP ETSI-op Encoder
* Build Aug 22 2005, 17:08:03
* Use of this product is permitted exclusively during the
* course of 3GPP codec evaluation.
* (c) 2000 - 2005 Coding Technologies
* (c) 1999 Fraunhofer Gesellschaft IIS-A
*
*************************************************************


configuration not supported

for testing, i have to force 48000hz output in wav (with wave_out plugin in winamp)
i must say: what a superb audio cd quality at 48kbps, excellent codec!!!!!!

dimzon
26th August 2005, 12:08
I really want to test FX's quality, but I keep running into this strange problem :eek:!
try previous build (look @ link in my first post)

dimzon
26th August 2005, 12:10
2everyone
I have really bad ears so it's i can't tell about quality. Can anybody produce blind test and public results?

DeathTheSheep
26th August 2005, 20:12
try previous build (look @ link in my first post)

Unfortunately, I believe the previous encoder used only floating point instructions. If you become aware of any other versions of the fixed point encoder, drop me a message ;)! Thanks much.

Can anybody produce blind test and public results?

Well, I don't have the best of hearing either, but I'm told that any difference can be very easily differentiated with the classical music pieces found on the hydrogenaudio site.

http://guruboolez.free.fr/

ggab
27th August 2005, 06:48
and, can i ask for a program that compare 2 different audio files, in particular, one is the input.wav, and the other is output.m4a converted again to wav (to avoid issues), so the program can compare the waves signals..... and tell us, if the 2nd file is identical or almost the same as the original ?

thnks

ggab
29th August 2005, 22:07
is there any program than keep the Id3 v2 tag in mp3s > while are converted to aac+v2 ?

tnks


PS: i wish the 44.1khz input wav files will be fixed soon :)

dimzon
30th August 2005, 11:35
PS: i wish the 44.1khz input wav files will be fixed soon :)
This encoder is tuned for 48000 encoding. I can't retune it for 44100. Just removing samplerate check is not a solution - you need to tune it for 44100.

Kurtnoise
30th August 2005, 13:20
I think AAC+ 48000 is better than AAC+ 44100 (because AAC+ nature)
Prove it.. :readrule:

Kurtnoise
30th August 2005, 13:25
is there any program than keep the Id3 v2 tag in mp3s > while are converted to aac+v2 ?
foobar for sure...but for the Nth time transcoding a lossy stream to a new lossy file hurts much more the quality.


i wish the 44.1khz input wav files will be fixed soon :)
I don't think so. It's not a bug. And don't forget this is a reference encoder...That means it is usefull only to make some comparisons with others encoders. Not to use for backup or anything else.

dimzon
30th August 2005, 13:29
And don't forget this is a reference encoder...That means it is usefull only to make some comparisons with others encoders. Not to use for backup or anything else.
Actually I use it for my DVD-Backups (stereo @ 64 kbps) - it sounds fine for me :D

Kurtnoise
30th August 2005, 13:34
Well...You're not afraid. ;) But If I was you, I'll wait the AAC-PS in the next Nero bundle.

dimzon
30th August 2005, 13:38
Well...You're not afraid. ;) But If I was you, I'll wait the AAC-PS in the next Nero bundle.
PS part working only @ bitrate below 32kbps. For bitrates 33kbps and above i't HE-AAC only (no PS). I have not good ears but some my frends tell me that referenced encoder sounds much better than Nero HE-AAC implementation even @64 kbps.

IgorC
30th August 2005, 17:25
New compile doesn't work for me (48/44.1 both wav sources aren't acceptable ). Only old does.

I don't need too low bitrate 32 kbit/s for my backups. But for 52 kbit/s (stereo) when PS is disabled it's still better than HE-AAC Nero to my ears. I prefer this HE-AAC-v2 audiocodec and not LC-AAC because of spotless diference for movie audio track, where mayor part of sound is voices,some background sounds(like shoots), background music.

Nero says that PS will be enable already in this year.

ggab
31st August 2005, 01:44
PS part working only @ bitrate below 32kbps. For bitrates 33kbps and above i't HE-AAC only (no PS). I have not good ears but some my frends tell me that referenced encoder sounds much better than Nero HE-AAC implementation even @64 kbps.
So, encoding with Besweet+belight will be the same results (as is it a Nero he-aac (aacenc32.dll) GUI), no?



well, i'm using previous v1.2 build (with no PS support) via dBpowerAMP Music Converter.
It converts 44.1khz OK, and the GUI helps a lot, but the important thing are:


1) To Prove, AAC+ 48000hz is better than AAC+ 44100 with/without PS enabled

2) Sound Quality with (for example at 64kbps) :
a) db_EnhAACPlus v1.2 encoder b) Fresh build of AACPlus v2

ggab
31st August 2005, 10:58
good news:

Winamp 5.1 Surround Edition is out :)

* New: CodingTechnologies AACPlus encoder!
The MP3 Killer!
Awesome at low bitrates too!


have a look:
http://img247.imageshack.us/img247/4605/dibujodd8ow.gif


question:
why the encoded RIP files are "MPEG-2 AAC" ? (well, there are .aac ....)
ATTACHED, the new demo.aac "DJ Mike Llama - Llama Whippin' Experience - Mix optimized for 5.1 surround. Stereo users will not get the full effect. - By CodingTechnologies aacPlus"

but if a demux a .mkv (h264 + he-aac) with MKVextractGUI, i get a "MPEG-4 AAC" in winamp file info box + id3 tag editor....

sasam
31st August 2005, 11:53
The download of new compile on yousendit expired. Can someone put it there again or mail me the encoder?
I really want to try it!
sasam(at)myrealbox.com
GGAB: Where did you get Winamp 5.1? I cant find it on winamp.com?

IgorC
31st August 2005, 17:32
Great quality for AACplusv2 Winamp 5.1. Yet another HE-AAC PS coder already enable.Nero has a reason to hurry with PS because there're already 2 HE-AACv2 codecs with superior quality of the sound.

dimzon
31st August 2005, 17:39
The download of new compile on yousendit expired. Can someone put it there again or mail me the encoder?
I really want to try it!
http://s41.yousendit.com/d.aspx?id=3EGQKQRFPQN9Y320KLRIZ2UOSE

sasam
31st August 2005, 19:16
Thanks...
Is it better than nero in mono at low bitrate, as I guess then it doesnt use PS.

dimzon
1st September 2005, 12:14
Great quality for AACplusv2 Winamp 5.1. Yet another HE-AAC PS coder already enable.Nero has a reason to hurry with PS because there're already 2 HE-AACv2 codecs with superior quality of the sound.
Seems like it's possible to write standalone commandline tool based on Winamp Plugin API to use enc_aacplus.dll :)

dimzon
1st September 2005, 18:26
Hi! I wrote standalone commandline tool based on Winamp Plugin API to use enc_aacplus.dll ;)
my source code is based onto The Winamp Transcoder (http://www.srcf.ucam.org/~wdhf2/transcoder/) plugin.

Get it here (http://s18.yousendit.com/d.aspx?id=1ZE251LLCRH3A2INNXWCU2121Z).

Note! Original enc_aacplus.dll not included - you need to download Winamp 5.1 to get it!

Note! You need at least in_wave.dll (input winamp plugin) to encode WAV file (not included too)

Note! Tool was tested with in_wave.dll and in_mp3.dll only (but other formats may work too)

Enjoy!

Nil Einne
1st September 2005, 23:08
Well...You're not afraid. ;) But If I was you, I'll wait the AAC-PS in the next Nero bundle.
What does it matter? The simple fact is if the quality is better then it is a better choice then Nero! Yes perhaps Nero will be better in the future but that is irrelevant to the now. Who cares what Nero will be like in the future? We live in the now.

BTW, it is a reference codec so even though perhaps the quality is not superb and it is probably very slow, the output should be extremelty standard compliant so there will be no compatibility problems in the future. If there is, it will almost certainly be the decoder that is at fault not the reference codec!

Rockaria
2nd September 2005, 04:30
The critical lacking feature with the reference codec/encoder is probably the nero's 5.1ch encoding/decoding support.
Also this current encoder isn't free from the input sampling rate unlike winamp 5.1's aacplus encoder which also still lacking the multi channel support.

You might be unable to settle down on this codec/encoder very long(can be obsolete very soon). :)

Kurtnoise
2nd September 2005, 08:52
What does it matter? The simple fact is if the quality is better then it is a better choice then Nero!
"better" is a wrong word in this board I recall you. In addition, of course that the reference encoder give us some good quality. Otherwhise it won't be a reference. ;) But as you know, its using is very limited. That's why I said waiting... to perform some comparisons.




btw, thanks Dimzon for this app.

Rockaria
2nd September 2005, 10:19
Finally the Winamp 5.1 official version (http://www.free-codecs.com/download/Winamp.htm) is out.

The aac, mp4(not aacplus), vorbis encoder(from the input plugin web page) seems to supports 5.1ch transcoding with the 'transcoder' plugin above.
For the correct play of the mp4(m4a), the plugins in the linked page need to be installed(the mp4 demuxer is failing maybe).
The ac3 input plugin in the linked site works well with 5.ch.
The flac encoder from the input plugin web page is failing in transcoding 5.1ch(not tested 2ch).
Also the recent 0.0001 version of in_dts can play the dts(wav) but in stereo downmixed mode...

Overall, it looks like a good competetion to foobar2k, vlc, ffdshow, ac3filter including other source dsfilters(except dts...).

Btw, what I get by the forum rule not to use the 'best ...' is limited to the ambiguous questions such as 'what is the best music player?'.
If the situation is very clear and specific, I think we can use it like 'what is the best quality commandline aac encoder to encode to aac sbr+ps for 48khz sources for 2ch music streaming purpose currently?' :)

dimzon
2nd September 2005, 16:13
New version is here (http://s51.yousendit.com/d.aspx?id=1T303I215G94R18L1ZB46SMIAB)

+ rewritten from scrach
+ improved stability
+ more frendly parameters
- no more in_*.dll support - now you can encode only PCM WAV

Enjoy!

PS. I'm thinking about beSweet integration :)

Jibreil
2nd September 2005, 16:41
Thanks a million! This is something I have been looking forward to for a long time.

Does iPod accept AAC Plus ?

bond
2nd September 2005, 16:58
Does iPod accept AAC Plus ?with low quality output, yes

Jibreil
2nd September 2005, 17:20
with low quality output, yes

:thanks:

Could you please elaborate by low-quality output?

dimzon
2nd September 2005, 17:22
:thanks:

Could you please elaborate by low-quality output?
iPod does'nt understand HE and PS part. So your files wil be played as half-samplerate AAC-LC
:search: :readfaq:

bond
2nd September 2005, 17:42
iPod does'nt understand HE and PS part. So your files wil be played as half-samplerate AAC-LChalf samplerate with he-aac and half samplerate + mono with ps-aac

Jibreil
2nd September 2005, 22:03
Thanks fellas

Hyper Shinchan
3rd September 2005, 17:05
Well, at least nero announced that they will add AAC Plus v2 support in this year, I'm wondering when Apple will add support for this......
anyway, actually what are the decoder that can understand AAC+HE+PS?
EDIT: anyway, tnx for all your useful build and tools, Dimzon. I'm living an hard time waiting for Nero AAC Plus 2.

ggab
3rd September 2005, 22:30
this one (in_mp4 v2.1beta plugin for winamp)
http://rarewares.org/dancer/dancer.php?f=10

Hyper Shinchan
4th September 2005, 17:49
Thank you, I have noticed this plugin when I used the link for Winamp 5.1 (but before i posted).
With the coreaac the PS part is ignored or not?

bond
4th September 2005, 22:43
With the coreaac the PS part is ignored or not?not ignored

Jibreil
5th September 2005, 10:54
Dimzon: Im really looking forward for the following:


To implement an interface or mux file support where we can load Six Mono WAV's and encoder with aacPlus v2 for a 5.1 channel output.
Encoder to support 6 channels instead of just 5.
Encoder to support Wave Ext. Format Header

Jibreil
5th September 2005, 11:02
Well, at least nero announced that they will add AAC Plus v2 support in this year, I'm wondering when Apple will add support for this......
anyway, actually what are the decoder that can understand AAC+HE+PS?
EDIT: anyway, tnx for all your useful build and tools, Dimzon. I'm living an hard time waiting for Nero AAC Plus 2.

If you've downloaded WinAmp 5.1. Go to the Plugins folder, you will find the in_mp4.dll file there.

bond
5th September 2005, 13:30
very interesting quality comparison between helix, codingtech and nero he-/ps-aac:

http://forum.doom9.org/showthread.php?p=707673#post707673
http://www.hydrogenaudio.org/forums/index.php?showtopic=36868

bond
5th September 2005, 17:24
Sources (http://www.3gpp.org/ftp/Specs/html-info/26410.htm)I've developed a CLI codec for Enhanced AACPlus, aka AACPlus v2, aka AAC+SBR+PS using the reference encoder sourcecode provided by Coding Technologies.ok so it seems coding technologies released the sourcecode for their ps-aac encoder as opensource for the 3gpp reference encoder here:
http://www.3gpp.org/ftp/Specs/html-info/26410.htm

but now my questions:
- is it sure that the 3gpp reference is from coding technologies?
- is the 3gpp reference the same or a different encoder than the one coding technologies sells?

would be happy if someone knows an answer to this :)

IF the 3gpp code and the one coding technologies sells is the same, why do we need to mess around with winamp and stuff like that?
simply pack the code into a nice standalone commandline encoder