New AAC LC-Main Profile Encoder / Decoder Options

[wkwai]

Hi,


I have completed my AAC LC/Main Profile MPEG4 Encoder / Decoder after 3 years..


1. The Encoder isn't optimized and currently running at 0.9x on a Celeron 650 MHz..

2. The Encoder can only take in 16 bits mono / stereo audio at 44.1khz.. Other options not supported.

3. The Decoder is fully conformance LC / Main profiles except that it can't handle 7.1 audio and CCE elements not implemented..

4. The Wav output of the Decoder is always mapped to the channel configurations of MS WAVE EXTENSIBLE format..


Please be free to try out this latest encoder..


wkwai

Attached File(s)

 aac_zip.zip ( 180.74K ) Number of downloads: 1352

 

[The Link]

Seems to work fine, only the progress indicator is broken here (goes forth and back quickly while encoding one file).
The resulting file (96 kbps/ch; mp4) sounds fine to me and I couldn't find any obvious flaws (for now).

Two questions:
1. Did I realize correctly that there's no vbr mode implemented yet?
2. Should "intensity stereo" also be used at higher (>64 kbps/ch) Bitrates?

Regards,
The Link

[wkwai]

Seems  to work fine, only the progress indicator is broken here (goes forth and back quickly while encoding one file).
The resulting file (96 kbps/ch; mp4) sounds fine to me and I couldn't find any obvious flaws (for now).

Two questions:
1. Did I realize correctly that there's no vbr mode implemented yet?
2. Should "intensity stereo" also be used at higher (>64 kbps/ch) Bitrates?

Regards,
The Link

Yes, it is constant bitrate.. but I think at 96 kbps, there is still some difficulties with some clips.. those with a lot of fast attacks.. Most of the testing was done at 128 kbps..

Actually the intensity stereo is automatically switching.. It is not permanently on..


wkwai

This post has been edited by wkwai: Aug 19 2004, 10:29

[The Link]

Actually the intensity stereo is automatically switching.. It is not permanently on..

That's good to know, thanks!

[jido]

The archive contains a single file, AAC_PROJ.EXE. Is that the encoder/decoder? Do you plan to release the source, or provide a MacOS X or Linux binary?

Thanks.

[Liquid_Predator]

Nice! I love to see new people developping AAC encoders/decoders!

Keep up the good work!

[wkwai]

The archive contains a single file, AAC_PROJ.EXE. Is that the encoder/decoder? Do you plan to release the source, or provide a MacOS X or Linux binary?

Thanks.

Yes the exe contain both the encoder and decoder.. It is just a demo program.. and GUI is NOT my area of specialization..

I don't really know if I would have the time for that .. Maybe..

wkwai

This post has been edited by wkwai: Aug 20 2004, 09:32

[mario620]

wkwai,

Are you going to include higher bit rates (320) in your next release?

[wkwai]

wkwai,

Are you going to include higher bit rates (320) in your next release?

I think it is not necessary to go that high.. I think not..

[jido]

The archive contains a single file, AAC_PROJ.EXE. Is that the encoder/decoder? Do you plan to release the source, or provide a MacOS X or Linux binary?

Thanks.

Yes the exe contain both the encoder and decoder.. It is just a demo program.. and GUI is NOT my area of specialization..

I don't really know if I would have the time for that .. Maybe..

wkwai

If it is standard C or C++, it should compile on Linux and [possibly] MacOS X without changes.

I don't understand the comment about the GUI? Both platforms have a command-line interface, something similar to a MSDOS window.

[wkwai]

The archive contains a single file, AAC_PROJ.EXE. Is that the encoder/decoder? Do you plan to release the source, or provide a MacOS X or Linux binary?

Thanks.

Yes the exe contain both the encoder and decoder.. It is just a demo program.. and GUI is NOT my area of specialization..

I don't really know if I would have the time for that .. Maybe..

wkwai

If it is standard C or C++, it should compile on Linux and [possibly] MacOS X without changes.

I don't understand the comment about the GUI? Both platforms have a command-line interface, something similar to a MSDOS window.

The problem is that.. the GUI is based on MFCs.. which is designed to work on the windows platform.. For other OS, there would be different windowing capabilities and my MFC based GUI won't be compatible with those OS..

Another problem is that my AAC codes are "integrated" into the GUI codes..

Anyway, I won't be porting the codes to other OS...

After all.. this is just a demo program..

wkwai

[enry2k]

Congratulations for your work. I guess you are an esperienced programmer! Where did you obtain all the technical data needed?
Keep working on it, maybe one day it will become the new "Lame" of AAC. Adding SBR High Efficiency AAC mode would be great!

Greetings

Enrico

[wkwai]

Congratulations for your work. I guess you are an esperienced programmer! Where did you obtain all the technical data needed?
Keep working on it, maybe one day it will become the new "Lame" of AAC. Adding SBR High Efficiency AAC mode would be great!

Greetings

Enrico

You can find all you need over the internet..

wkwai

[enry2k]

Thank you.
I hope you will keep developing your project.
Why don't you create a standalone utility to batch convert aac files from any format (.aac adts, .mp4 m4a) to another with a single windows GUI executable file?

Regards

Enrico

[wkwai]

Thank you.
I hope you will keep developing your project.
Why don't you create a standalone utility to batch convert aac files from any format (.aac adts, .mp4 m4a) to another with a single windows GUI executable file?

Regards

Enrico

I don't think I will be continuing with the project because I will be busy with my Master Degree programme.. Maybe I will continue with it after the completion of my studies.. However, a lot depends on whether I am distracted or not.. If I am distracted in my studies then, it would take a longer time to complete...

wkwai

[MugFunky]

good work, wkwai!

decoded fine in everything i tried it in (i don't have an iPod...), and sound was good, considering it's a demonstration and hasn't had the bejeezus tuned out of it.

it seems like you've been working on this encoder forever. nice to see it's finished and sounding nice (not too slow on my machine, p3 733. i never expect miracles with a machine that's nearly 5 years old).

[kotrtim]

I tried 2 samples
1.One from "Simple Plan" which is Punk Rock
2.pop music

all AAC codec performed better in sample 2 than sample 1
I'm too lazy to ABX higher so i just use CBR 96kbps for this test, which is very easy!

I'm shocked, Quicktime 6 used to be very bad below 112kbps
QT 6.5.1 is ok, and appears to be the best 96kbps AAC encoder, at least for these 2 samples, QT has the least pre-echo......wkwai's perfomance in sample 1(rock) is as good as QT , i cannot decide which is better..
for sample 2, QT is obviously a bit better than wkwai's

if u guys want the sample, i can upload them

this is my rating, from best to worst (for these 2 samples only!)
1. QT 6.5.1
2. wkwai's AAC
3. Nero
4. WMA 9.1 standard

[rjamorim]

That's interesting. So we might have another high-quality AAC encoding option.

[kotrtim]

if anyone is interested,

this is the sample 2 i mentioned

sample 2

[dand]

That's interesting. So we might have another high-quality AAC encoding option.

i agree that wkw encoder is surprisingly promising, but let's not base our conclusions on just a few test samples. rob, i guess it's time for another (wkw centered) aac listening test!?

[wkwai]

That's interesting. So we might have another high-quality AAC encoding option.

i agree that wkw encoder is surprisingly promising, but let's not base our conclusions on just a few test samples. rob, i guess it's time for another (wkw centered) aac listening test!?

Are you really sure of it?? I mean, I implemented the spreading function of the psychoacoustic based on Yan's proposal of not symetrical slopes.. which I now think is wrong..

I have another better encoder.. which I will upload another time..

wkwai

[Enig123]

We are all expecting for your new (better) encoder.

Good work, wkwai. Go on coding...

BTW, can you obtain some information of "Yan's proposal", url? or upload a document?

Edit: Oh, I just found the information I want in this forum. Now I'm reading...

This post has been edited by Enig123: Sep 25 2004, 14:44

[rjamorim]

i agree that wkw encoder is surprisingly promising, but let's not base our conclusions on just a few test samples.

That's why I wrote "might"

rob, i guess it's time for another (wkw centered) aac listening test!?

I am retired from test conducing. But I would be happy to help people interested in conducing.

I think a new AAC test will be justifiable once the next QuickTime is released (if it really comes with VBR) and the much-postponed Nero encoder update is launched.

[Enig123]

It has been reported that Real's aac coder suffers from the lowpass (15kHz IIRC).

I wonder if this aac coder can benifit from higher lowpass, e.g., 18kHz (there's an option in the interface). It is known that aac specification didn't bother with sfb12 problem like with mp3's, so the quality should not suffer from higher lowpass.

Correct me if I'm wrong.

[Enig123]

Are you really sure of it?? I mean, I implemented the spreading function of the psychoacoustic based on Yan's proposal of not symetrical slopes.. which I now think is wrong..

I have another better encoder.. which I will upload another time..

wkwai

I'm interested in what will happen if the wrong spreading function is adopted in the encoder side. I read the threat about Yan's proposal and it suffers the sound quality to use the wrong spreading function and there's some people prove the conclusion by blind test. But it seems this coder doesn't suffer much from this. That suprise me.

Maby it's because the output stream is aac not mp3??