Winamp FhG AAC encoder with gapless fix, Updates from the 5.62 release Options

[Gainless]

The first one is pretty clear, the second one is already very quiet, and after that I hear nothing at all.

I guess that's not really good, isn't it?

That's a common misconception here on HA. You listened at normal levels (i.e. not cranking up the volume), so your performance is perfectly normal, maybe even above average for your age. At normal levels I get the same result as you. Bottom line: you might have heard artifacts below 16 kHz, so I'm not going to put more bits into high frequencies.

Is this "issue" a show-stopper for you? Apparently this was very hard to ABX. Can you live with the encoder's current performance on this item? Or are you really looking for utmost transparency?

Chris

Good question. I'm a kind of lossless-junkie so I would more tend to "yes", but of course these kind of issues don't turn the encoder to "completely awful" for me. So in conclusion it would be cool if it got "fixed" as it would make the encoder pretty flawless, but I wouldn't throw Winamp off the computer if not.

This post has been edited by Gainless: Jan 15 2012, 15:08

[IgorC]

Gainless,

It would makes sense to see how other AAC encoders perform on this sample to see if it's quite normal behavior for the entire AAC format or it's only one particular encoder.
Let say to see how perform Apple and Nero. Previously Replaygain should be applied for Apple encoder (or to all encoders as well) as it changes volume noticeably.

[Gainless]

Gainless,

It would makes sense to see how other AAC encoders perform on this sample to see if it's quite normal behavior for the entire AAC format or it's only one particular encoder.
Let say to see how perform Apple and Nero. Previously Replaygain should be applied for Apple encoder (or to all encoders as well) as it changes volume noticeably.

Judging from the ABX'ing I've done yet it's even more noticeable with Quicktime AAC at TVBR 95 (has 13 kbps less on average though), but can't prove it already, I'll upload a log later.

This post has been edited by Gainless: Jan 15 2012, 18:38

[IgorC]

Then it's not only FhG.

Can You try CVBR, Nero and FhG all together in ABC/HR? With this http://ff123.net/abchr/abchr.html

This post has been edited by IgorC: Jan 15 2012, 20:21

[Gainless]

Then it's not only FhG.

Can You try CVBR, Nero and FhG all together in ABC/HR? With this http://ff123.net/abchr/abchr.html

Should be no problem at all, but this will take very long with at least 30 listenings. Btw, have you already tried to ABX the sample? It would be great to hear some other views on it.

[IgorC]

Btw, have you already tried to ABX the sample? It would be great to hear some other views on it.

Sorry, I have no time right now. Will try it later. I mention the ABC/HR because it can be useful for Chris to see if there is any other AAC encoder that can handle this sample better, hence there will be still room for improvement.

[Pete7874]

Where can I find a list of available parameters that can be used with this FhG AAC encoder?

I'm looking for a parameter that would allow me to automatically create a "Tool" tag for each file being converted with FhG. This "Tool" tag just needs to say "FhG AAC version x.x.x.x" so that I know what encoder was used to create a particular file. Nero and LAME (and probably many other encoders) add this tag by default.

Thanks!

This post has been edited by Pete7874: Jan 20 2012, 21:46

[C.R.Helmrich]

Fraunhofer's AAC encoder does not add such a "tool" tag at the moment, and I don't think Winamp offers a switch for this either. But I'll keep in mind that there's interest in such metadata.

Chris

[Pete7874]

Thank you, Chris!

[matt the cat]

Fraunhofer's AAC encoder does not add such a "tool" tag at the moment, and I don't think Winamp offers a switch for this either. But I'll keep in mind that there's interest in such metadata.

Chris

Yes, this would be really nice. Although I would like to see in addition to the encoder and version, the parametres used to create the file, just like Nero AAC and Quicktime does.

[benski]

I should be able to add this in the next version.

[Jebus]

Hey,

Late to the party here... I have 2 questions about this DLL:

1) Is there any API documentation anywhere?
2) Is there a 64-bit version somewhere?

(I've emailed FHG licensing too)

[Jebus]

They got back to me that licensing their object code would be "about the price of an expensive car. Still interested?". That's kind of a dick answer hey? I was requesting specifics.

[Gainless]

Little update for my ABXing claims:
I don't really know why, but the more I ABX, the worse results I get. I've tried the sample with Nero, Quicktime and FhG today again, and couldn't achieve anything far below 50%. Seems like the difference to the lossy files is not significant enough to be noticeable for me anymore. I'll try it later again, but I'm not too optimistic to get better results...

This post has been edited by Gainless: Jan 31 2012, 16:11

[LigH]

Quality-wise, we are not planning any huge updates, at least not for stereo and mono. But if there is a significant interest in other VBR modes, we will consider adding those.

Certainly yes. Different people will have a different threshold of bitrate and quality, due to most different reasons. May it be just to "feel safe". Your quality steps are quite coarse, so there may be people (like me) who feel like VBR preset 5 "wasting size", but preset 4 "sparing quality".

As an example, one may compare with other formats, like MP3 or Ogg Vorbis: MP3 should have around 192 kbps, preferably VBR (e.g. lame -V 2), and because it is less efficient than the current generation of encoding algorithms, AAC with a comparable quality may need only 75-80% of the MP3 bitrate, or about the same as Ogg Vorbis at -q 5, so the target for a VBR AAC encode would possibly be around 160 kbps.

QuickTime uses discrete steps too, but they are finer. Their --tvbr [<=]91 seems to be similar to your --vbr 5, but the next lower step, --tvbr [<=]82, appears to give results quite close to oggenc2 -q 5 or ~80% of lame -V 2.

[C.R.Helmrich]

Quality-wise, we are not planning any huge updates, at least not for stereo and mono. But if there is a significant interest in other VBR modes, we will consider adding those.

Certainly yes. Different people will have a different threshold of bitrate and quality, due to most different reasons. May it be just to "feel safe". Your quality steps are quite coarse, so there may be people (like me) who feel like VBR preset 5 "wasting size", but preset 4 "sparing quality".

As mentioned here, it seems Winamp 5.63 has been released, with an updated AAC encoder version which now includes a VBR mode 6 (roughly 256 kbit/s stereo on average). VBR mode 5 should now have a slightly lower average bit-rate than in older versions (closer to 192 kbit/s stereo).

Chris

[punkrockdude]

New: [enc_vorbis] Ogg Vorbis Encoder (based on aoTuV)

I like this!

[LigH]

As mentioned here, it seems Winamp 5.63 has been released, with an updated AAC encoder version which now includes a VBR mode 6 (roughly 256 kbit/s stereo on average). VBR mode 5 should now have a slightly lower average bit-rate than in older versions (closer to 192 kbit/s stereo).

Okay, but...

The gap between mode 4 (~128 kbps) and 5 (~192 kbps) is too large for my taste. I'm hoping for a mode 4.5 (~160 kbps).

[nu774]

Is this the same implementation as is used in Winamp?

Fraunhofer FDK AAC code from Android.

[darkbyte]

Any news about HD-AAC encoding? Are Winamp devs still considering to adding it or they dropped it because of the lack of interest?