I am new to aac so I need to ask some questions:

i) aac and mp4 apparently have close ties, but which is better for archiving (medium to high bitrates)?
I've heard of differences in encoding time and loading time. Which are the differences exactly?
ii) Is aac_enc v2.15 the best aac encoder? I downloaded the zip file but the accompanying txt
told me that it is v2, and not v2.15... What's up with that?
iii) I encoded a song with EAC and aac_enc v2.15. It took ages to encode. When I played it in Winamp 2.81
it took a couple of seconds for it to start, and 2 seconds before the end of the song an error message
appeared reading "unable to find ADTS syncword". This can't be good... I'm using the in_faad and
in_mp4 plugins which john33 recommended in the first message of this thread. What can I do about
this problem?
iv) DSP plugins like DFX work fine with ogg and mpc. Does this also count for aac and mp4?
v) Where can I find an mp4 encoder?
vi) With mpc out there, is there any reason to use mp4 or aac?
vii) What is faad?

i) The quality is the same (raw data inside he files is exactly the same). MP4 ususally carries a bit less sideinfo so it is usually (a little bit) smaller. MP4 files can be loaded easier because the sideinfo it carries has some more info on song length etc. For AAC the whole file has to be read to know the exact length.

iii) The error you got clearly shows that you have version 2 and not 2.15.

iv) DSP plugins work with both the plugins you mentioned. in_mp4 can also play AAC files and doesn't take as long to load the files. Problem is that it does not show the file length and does not allow seeking in the aac file.

v) Quicktime 6 Pro
and Nero will have one soon.

vi) yes, if you want future support by almost all players. Hardware support, audio with MPEG-4 video, you name it....

vii) it's the name of an AAC decoder (http://www.audiocoding.com/)

Menno

Rarewares hosts the latest Psytel AACenc 2.15. Probably the best easily available AAC/MP4 encoder will be the Nero MP4/aac encoder plugin. Psytel aacenc is after the Nero plugin release at least quality wise obsolete.

Could you give me link to aac_enc v2.15?

Rarewares: http://www.inf.ufpr.br/~rja00/

Uhmm... this page contains only news. which file am I meant to download?
I've already downloaded aacenc_v215.zip from this site, but this was the exact encoder that
gave me the "missing ADTS syncword" problem...

You can use your current AACENC v2.15 as an mp4 encoder (once you have it working that is!)


Get the AACENC.EXE from Rarewares
Get the MP4CREATOR.EXE from there too (you have to rename it from mp4creator60.exe)

Then to save you typing in loads of commands, get the great frontend from Speek here

Then put the 2 exe files in the same directory as the frontend, select aacenc_mpeg4 as your encoder, mp4 as the encode type (not raw aac), and put your settings (-standard or -streaming, etc..) in the box underneath It should all hopefully run then!

You might not get the ADTS error using the mpeg4 version, I'm not sure though..


AAC is better at low bitrate encoding, well, anything under 128kbs
And it is, or should be a much more popular format soon, having portable players and the like, which MPC is heh, having trouble getting 1 of!

Thanx a lot 4 your advice, just one question though: what do I need MP4CREATOR.EXE for?
I mean, you didn't tell me what to do with it besides putting in the same directory as the frontend...

I don't think it can be explained much more clearly than Mac did. You use MP4creator to make .mp4 container files out of .aac files..

Could u explain to me what this means?
Is MP4creator used by the frontend?



U mean the mpeg4 version of what? Of the winamp plugin?

Yes, it's used by the frontend. mp4-container (.mp4 file) is the container for industry standard MPEG4 audio/video. When you encode with Psytel, you get by default MPEG2-AAC LC .aac -file. Then you use MP4creator to wrap the file into .mp4 container format - and it becomes MPEG-4 AAC LC mp4 -stream, which you can play for example with in_mp4.dll WinAmp plugin.

After Mac's helpful (and most importantly LINK-full ) post I gave AAC a shot and I like what I hear.
Taking into account the future support of the .mp4 container, I would be grateful if you could point me to a thread discussing AAC vs. MPC at archival level. I suspect who the winner is, but I would like to follow such a discussion, if there has been one of course!
TIA

Uhhh, i got some questions too....
What is the difference between AAC low complexity and AAC main in the -profile option... ?! I guess that one optimize the sound... But... i don't know...
And are both compatible with EXP 401 phillips player?! Thanks....

mpc gives transparent quality at 192kbps, while aac is 100% transparent at 224 IMO. BUt i stay away from MPC since it lacks of any hardware support, on the other hand phillips expanium is already here with full AAC support. Even though the manual syas 160kbps and below, im pretty sure some peeps around here tried extreme preset and it worked flawless

Just like the name "low complexity" implies, this profile is not as demanding for encoders and decoders, so that time, memory and CPU usage are not critical at least for average computers. The "main" profile uses enhanced techniques/tools for a better result, but it seems to be in a quite early stage of development, because it makes things worse than the LC profile in PsyTEL AACEnc yet. One part of this profile are prediction methods that analyze the signal for noise and tonal components and try to substitute them via the decoder. LTP (Long Term Prediction) nevertheless is not a part of the main profile, but at least in PsyTEL AACEnc a profile of its own.



No, the Expanium only plays MPEG-2 AAC LC files, i.e. not the main profile or LTP. It also only recognizes the ending *.aac, not *.mp4, so you have to use a compatible encoder to produce playable files.

And it can't fully decode aacPlus of course (your other question), because it can't do SBR (Spectral Band Replication), also a new part of the MPEG-4 standard. But it will play aacPlus files backwards compatible (if they have the ending *.aac), which means that the upper part of the frequency range will be left out during playback, just like mp3PRO files decoded with a normal MP3 player (hardware or software).

MPC will usually give transparent results at lower bitrates than this (~160 or maybe even less) on most samples and for most people.



No lossy codec is 100% transparent yet.

ok sorry about my n00bism dibrom, mpc sounds good, but heck it will have hard time getting hardware support. Of course i didnt mean 224kbps AAc is perfect, but i cant abx from the real source even on problem samples and also i dont have any fatigue unlike mp3. Sorry, but well im here to learn

yet?

Is aac/mp4 VBR? I'm asking because when I play an mp4 file with Winamp the bitrate does not
change... Or is the average bit rate displayed?

Something else: low complexity means low quality?
is there an other mode, and how does one choose it?

its vbr, and low complexity profile needs relatively low processing power to encode/decode aac files, so main profile would need more horsepower.

No, absolutely not Difference between best AAC Main and AAC LC implementation is probably too small for most average users to be detected even.

Prediction used in Main profile has much more disadvantages than small advantages it offers (for tonal signals), and it is practically "obsolete"

MPEG-4 AAC LTP prediction is less complex than Main profile, with same quality - but it is also not widely used (almost all hardware devices are LC only)

New "high efficiency" MPEG-4 profile will define MPEG-4 AAC LC + SBR

I've heard of the use of SBR in mp3pro...
Will it also be incorporated in aac/mpeg4? I thought it was suitable for low bitrates only...
Could u explain in layman's terms whay SBR offers?

Somewhat irrelevant question: why is mp3pro limited to 96 kbps?

Yeah, SBR will be in AAC+ although I'm not sure if it's the same exact tech that's used in mp3PRO.
Pretty much yeah. AAC+ is meant for lower bitrates.

http://www.hydrogenaudio.org/forums/index....2736#entry26972
It's not limited to 96kbps.

I heard something like "when DVD based portable players come MPEG-4 will be automatically supported" somewhere in hydrogenaudio forums. Is it the case? Why?

And among the AAC encoders (Quicktime Pro, PsyTel, the new Nero plug-in), I assume the Nero plug-in will be usually the choice of AAC users here. Is there any way we can actually use it with EAC? or can we use EAC it with Quicktime Pro encoders?

Alp

yeah, just rip the tracks using EAC as usual (i dont stop until evey track has 100% quality), and nero has an option of Tools/File Encoding...

Right now you can use mp3/PRO, vqif and other stuff. We are waiting for the next version that will incorporate psytel AAC 2.40.

Hmm, what about the tags then? Too much of a hassle. I actually like Vorbis quite a lot. But future hardware support is of utmost importance.
If I iRiver comes up with Ogg Vorbis support as they promised a while ago, I don't see a personal reason to go with AAC.

Alp

I don't think that if u persist, u'll get 100% quality... If a track is ripped at a quality of 99,7%
then 99,7% it shall be... Or not?

of course i tweak settings, i lower read spead and switch c2 on-off until i get 100%, not a single cd ive tried has been impossible to rip with 100% quality. I have a plextor 24x burner.

Atici, i dont know if you read the thread about monty's comments on high bitrates with vorbis, to sume it up, they will emphatize on 128 and below. Its good if you want good quality ratios, but audiophiles wouldnt like to loose any quality at all. MPC is quite good at 160+, but i will stay with AAC since it has already got hardware support and 192+ bitrates are awesome (much better than vorbis).

On the tags issue, you can still use Tagger which supports lots of formats, but i wouldnt exchange wimpy tags for audio codec quality.

If you agree with me cool, if not, well, thats why we have all this options

http://pages.cthome.net/homepage/eac/eacfaq.htm

Q. After I've extracted my tracks the status message reports less than 100% track quality but I hear no glitches, what's the deal?

A. The track quality is the ratio between read operations that are needed to perform the task and real total read operations performed on that track... 100% - Only the needed read operations were performed.... If any rereading is performed it will lower track quality. Even if you have track quality less than 100% you are still getting an accurate copy. Errors and glitches only occur if EAC reports a suspicious position or a read/sync error.

ACK!!

well, ill still play it safe, dont want "rereads"..

I have small note regarding AAC hardware support:

AAC has been ported on these DSP and VLSI platforms:

http://www.iis.fraunhofer.de/amm/products/..._dsp/index.html

and this: http://www.tensilica.com/html/mpeg_2_decoder.html

and this: http://www.corepool.com/products/databook/.../fhg_aacdec.pdf

and this: http://www.micronas.com/products/overview/...media/index.php

and this: http://www.semiconductors.philips.com/acro...ecoder-001F.pdf

and this: http://www.arm.com/armtech.nsf/iwpList73/C...le=IP_Solutions


So, most important are here (TI, Motorola, Analog Devices, ARM, TriMedia, ARC and couple of chips - like Crystal, Micronas, etc...)

Hey, it runs on Mot's DSP56300 @ 25MHz

Great thx hans-jürgen... I've learn a lot...
So i'll continue to encode at low complexity
And now i understand why a .aac file in my EXP players didn't plays... that was just a test but ...
bye.

Err... re-reads just mean that EAC is making sure that data it was uncertain about is accurate. Like was said, if EAC does not report any suspicious positions, there is no reason to worry. At all. It just means that the CD was a little harder than average to rip. Heh, I think once I had a perfect rip at 98.3% quality. Might have been 99.3%, though...

BTW, is it possible to get a suspicious position in secure mode without getting a read or sync error?

Sometimes I get a test and copy both at say 99.9% but with different CRCs. How can they both be accurate if they are different?

Yep - Just got one

Maybe you could hear a reason if you'd try...

You're welcome... If you want to know more about AAC/MP4, you could also visit http://www.audiocoding.com and read through the forum, the Wiki (a user-editable knowledge base), the link section and even the industry news on the main page that sometimes are very interesting to read.



By the way, one of the links Ivan provided shows the exact capabilities of the Philips Expanium AAC decoder. This is the main part from the PDF file:

AdecAAC - Description:

The MPEG2 AAC Decoder library can be used in any product using a 16/32-bit TriMedia VLIW processor. It can handle ADIF as well as ADTS bitstream types. Raw AAC bitstreams can also be decoded, provided that the profile and the sampling rate are provided. Currently the low complexity profile is implemented.

The AAC decoder will work with all default channel configurations, as defined in the AAC specification. It will attempt to decode any reasonable nonstandard channel configuration as well, and match it to the output format of the device. Downmix of the 3/2 standard configurations to the stereo output format is an option. Only one audio program will be decoded and played. The output sample always has 16-bit precision, but both 16 and 32-bit output formats are supported.

A special feature of the decoder is its second output which can be used to send a stereo audio to a VCR, or encoded 1937 data to an external decoder.

Note that the MPEG 4 version of AAC is also under development.

Applications:

• BS Digital Japanese TV system
• Set Top Boxes for internet audio

Documentation:

A detailed document describing the API and the internal behaviour of the component is available.

Is this figure comparable to new Ahead AAC MP4? I’m interested in the comparison of MP4 to MPC, simply in terms of quality. Anybody with golden ears have an opinion that they can give me?

Are both encoders considered to be transparent even with their “standard” profiles? In this case, I would assume that quality is about the same for both, and choosing one would only depend on your preference for hardware support in the future, ect…

And how can I register the Ahead encoder, I see no MP4 link on the web site….?

thanks

I don't know if I fall under this category in your opinion, but I just compared PsyTEL v2.15 -normal with Musepack v1.14 -standard on my one and only test file for half an hour and I can't hear a difference except for especially the last of the three cymbal crashes in the opera excerpt, which PsyTEL seems to "enhance" just a tiny little bit. If you prefer some ABX charts and/or more samples to be used, there will perhaps be a true "golden ear" who wants to provide this for you.



The verdict "transparent" can become a misunderstood fetish sometimes, because I don't think there is any lossy codec that won't ever be "knocked out" or ABXed by someone with some strange "codec killer samples" provided by guruboolez or others... At least not at reasonable bitrates that are far from the ones that lossless codecs produce, and that's why I like the way the EBU handles this dilemma for their professional listening tests. They use the term "indistinguishable from the original" if the statistical results from their comparisons reach a certain level of confidence. This even includes the possibility that there might have been one or more participants in the test who heard a slight difference on one sample, but they were the only ones and it was not a constant tendency through most of the samples. This refers to what you cited from Dibrom: "... on most samples and for most people".

So if you don't want to do your own listening tests, you're stuck with someone else's opinion, because there haven't been any valid comparisons between the codecs discussed in this forum at the bitrates in question yet, only single reports about this and that problem sample from different people with different tastes of music and different hardware.

By the way, if you now decide to encode all your future MP4 files with the -extreme preset (don't know if Ivan changed the name in Nero to something else) for "peace of mind" reasons maybe (which would have made a good preset name, too... ), you should first test if Winamp can play these files with the latest in_mp4.dll, because there is or was a bug with PsyTEL's converted AAC files with this preset.



Probably this will be possible if they decide to publically announce it on their website, but I don't know anything about this, only guessing...

Retail version will be out very soon - and it will have one bug fix

Extreme preset is called the same in NERO MP4 plug-in. There have been some tweakings to it, so the bit rate might not be the same as with aacenc 2.15 (someone could give it a try).

Which bug ?
Have someone notice strange behaviour with -ultra profile ? I have very low bitrate with it. I encode two tracks at -ultra : 119 kbps and 133 kbps.

Other settings :

· ENCODER QUALITY : HIGH
· AAC PROFILE : LC
No boxes were checked

It is related to mono encoding. Regarding ultra - I'll do a check on Monday.

I'll do some extreme and ultra encodes tonigth for some testing.

I don't know if I encounter decoding problems, but I found the new AAC/MP4 codec to be worse than the old PsyTEL 2.15 codec. I'm just talking about one sample, and I'm not trying to conclude anything with this first experience. But the new codec seems to have a really bad behaviour. In few words, the bitrate is low for a so difficult passage :

· normal = 140 kbps
· extreme = 170 kbps
· audiophile = 213 kbps
· ultra = 119 kbps

· ogg vorbis 1.0 -q5 = 201 kbps
· musepack 1.14 --standard = 222 kbps
[these values concern the whole track : 270 seconds approx.]

I said 'bad behaviour', not only for the average bitrate, but for the poor quality I obtain. -audiophile is easy to ABX, and maybe as an harpsichord lover I find -extreme really awfull, maybe worse than --alt-preset standard (the sample is a manifest failure for the lame preset). -ultra is unusable...
I haven't the time and the ears (I've the flu) to perform a good comparison, but i'm really sure that PsyTEL 2.15 better is.

I'm hosting the first seconds for some hours/days :

http://membres.lycos.fr/guruboolez/

"Retail version" sounds good, will it be a command line EXE or a DLL?

And the bugfix "related to mono encoding": does it solve the "channel coupling not yet implemented" error message I get with converted AACEnc -extreme files in Winamp? Or does this error not occur with Nero AAC/MP4 at all?



Did you also tweak the -normal preset, so that the comparison I made in this thread to Musepack --standard might also come out differently with Nero instead of PsyTEL? That very small difference on the last cymbal crash might be "cured" with a little lower cutoff perhaps, because 20.5 kHz is not necessary for most of the existing music in my opinion and a lower one could make things easier for audible lower frequencies around 16 kHz and the remaining artifacts in this region.

I just tried on some different tracks.
First, the details for the harpsichord track :



Retrogradation ? Extreme appears like normal, normal like streaming, etc...

Other samples.

I can't reproduce this. Can you send me a short sample file on which this happens?

Menno

Hmm - TMN, NMT and other psychoacoustic values werent changed across the presets - there has been many issues in tonality estimation - that should be corrected now, and new TNS - so this is probably why the bit rate is lower.

I'll test some clips tomorrow.

Ah yes,

For mono samples - demo version has flaw block-switching issue ! It has been corrected, but it will be out in the retail version, and in the next demo version of nero! I am sorry for the inconvinience, but somehow this mono bug managed to sneek during testing! I apologize for this! It seems that for mono samples all blocks are LONG (in 5.5.10 demo from few days)

I'll try to upload fixed trial binaries tomorrow somewhere - so mono testing could be done.

Note - retail (web) version is fixed (I found the issue before preparing the web version - which will become available soon)

Hmm, you already confirmed this bug a few days ago, either in this thread or somewhere else... But I'll send you this file (466 KB) of course, shall I use your Audiocoding.com address again? I encoded the c't reference.wav with AACENC_MPEG4.EXE -extreme and converted it to MP4 with in_mp4.dll (not sure anymore with which version, probably John33's one from Nov 2, 2002). Trying to play the file in Winamp 2.81 pops up a little message window saying "channel coupling not yet implemented". This does not happen with files encoded with -normal or other presets and converted to MP4.