NeroAacEnc and 24bit wav files.

[jonathonsunshine]

I use EAC3TO (2.5) to decode a Dolby Tru-HD stream to a 6 channel wav file and then feed that to NeroAACEnc.

EAC3TO says that the Tru-HD stream is 24bit, am I better off having EAC3TO down convert it to 16bit first? Or just feeding the 24bit wav to NeroAacEnc directly ?

Cheers

[menno]

No need to convert. The encoder takes floating point as input internally and AAC itself has no notion of bit depth. Better to input as much information as you can.

[jonathonsunshine]

Thank you :)

[Seraphic-]

Does it also accept 96kHz and encode it as 96kHz or convert to 48kHz?

[menno]

It encodes at 96kHz. Here I would say it's better to resample to 48kHz before encoding, especially when encoding at typical AAC bitrates.

[Seraphic-]

It encodes at 96kHz. Here I would say it's better to resample to 48kHz before encoding, especially when encoding at typical AAC bitrates.

Thanks. I have a few more questions if you have a chance.

1. When "Increase Volume automatically" is enabled, how much does it increase the volume of the overall input file? Searching around I found a few people say to disable it while others say leave it enabled.

2. For Variable Bitrate encoding, is moving the drag bar left or right higher quality (I.E Q-0 - Q-1)? Searching around the web resulted in posts saying two different things.

3. For 5.1 audio, would AAC be recommended for videos over internet distribution or might it be unsupported by majority of users?

[menno]

I can only comment on 3. the rest are eac3to options I guess?

Re 3: If your users are using mostly software players there should be no problem, most current hardware players with AAC support don't do more than 2 channels AFAIK.

[tebasuna51]

1. When "Increase Volume automatically" is enabled, how much does it increase the volume of the overall input file? Searching around I found a few people say to disable it while others say leave it enabled.

Maybe Megui option?
"Increase Volume automatically" can be:

1) Apply Dynamic Range Compression if possible. If source is an ac3 encoded with the Film Standard DRC the high volume is attenuated until -20 dB, and sounds below -31 dB (dialog reference) are amplified. See here (http://pages.sbcglobal.net/wilsondr/ddcompprof.gif).

2) Normalize at 100%. Now the high volume go another time to 0 dB and dialogs (-31 dB) go until -11 dB.

Apply the DRC is not recommended when transcode (only at play time when is needed) because lose quality.

2. For Variable Bitrate encoding, is moving the drag bar left or right higher quality (I.E Q-0 - Q-1)? Searching around the web resulted in posts saying two different things.

Drag bar? Maybe in BeLight or in BeHappy?
-Q 0.1 (poor quality)
-Q 0.99 (max quality)

[Seraphic-]

Thanks and yes, was talking about Megui.
(sorry, thought I put that in my other post)

[jonathonsunshine]

1. When "Increase Volume automatically" is enabled, how much does it increase the volume of the overall input file? Searching around I found a few people say to disable it while others say leave it enabled.

I believe that MeGUI uses "peak normalization" like as in AutoGK or the Normalize function in AviSynth or BeSweetGUI. It finds the loudest point in the the sound clip and raises or lowers the volume of the whole clip so that loudest point is as the set max volume.

For movies I think this sucks, cause the loudest point in a films sound track can sometimes be a single gunshot or even sometimes a sound that is so short that you don't even notice how loud it is (seriously, like a tiny pop in the rustling of leaves or something can sometimes be louder than the gun shot). The end result is that the overall sound level is very low. Also, (peak normalizers) can have trouble different frequencise, for example, a full scale sine wave at 1kHz sounds much louder than a full scale sine wave at 10kHz but the percieved volume, what you can actually hear, is totally different but peak normalizers can't tell the difference.

I recommend using MP3Gain for MP3, AacGain for mp4/m4a audio streams (despite the name it doesn't actually handle raw aac streams) and the creatively titled WAVEGain for wav. Those are all command line utilities but they all have GUI's you can use as well. They do not actually re-encode the file, they "alter" the file, they tell the file to play a little bit louder, so there is no loss in quality.

Those programs all use the replaygain method of picking the idel volume. I'd swear by them.

Unfortunatly, WavGain doesn't like wav files that arn't stereo but AACGain will handle 5.1 AAC files just fine. Replaygain was developed for music mp3s (on account that in your play list you prolly have many tracks that have large difference in the percieved volume) but I have been using it for years on any audio stream I commit to disc, (Ie, extracting, normalizing and then remuxing it to the video).

There is a fair wack of info about it here http://replaygain.hydrogenaudio.org/contents.html



-DISCLAIMER
I did some comparisons on different sound clips created with MeGUI, and it ISN'T using the same method as MP3Gain, but in all fairness I don't know precisley how MeGUI normalizes the sound but I would still swear by the Replaygain method.

[tebasuna51]

I believe that MeGUI uses "peak normalization" like as in AutoGK or the Normalize function in AviSynth...
Of course, is the unique method to preserve original quality.

I never can recommend other method for backup purpose.
Other question is if you want make a low quality copy to stream by Internet or play in low end audio equipment.

MP3Gain, AacGain, WAVEGain ...
They do not actually re-encode the file, they "alter" the file, they tell the file to play a little bit louder, so there is no loss in quality.

Sorry but isn't always true. Like apply the DRC to an ac3 these methods lose quality.

[smok3]

Mp3gain is 100% reversable, wavegain is not, dunno about AACgain. replaygain is not similar to DRC in any way.

[Seraphic-]

I've been doing some more testing with Nero AAC (using MeGUI), DTS, and AC3 5.1 and found AC3 to honestly sound the least desirable for the audio I'm working with.
DTS sounded much more "vibrant" in comparison. AAC sounds as good if not better then DTS, but the playback volume is so much louder then DTS.

Here is a sample pack with my encodes.
Comes with raw WAV, 5.1 AC3, 5.1 DTS, 5.1 Nero AAC (one increase volume enabled and one disabled)
Sample Pack (http://www.temp.seraphicgate.com/audiosamples.rar)

I used these settings for AC3: (same but changed to 640kbps)
http://forum.doom9.org/showpost.php?p=907625&postcount=2

DTS settings were at 1.536 mb/s datarate with attenuate Rear Channels 3dB disabled

AAC used VBR at Q-0.99 and improve accuracy enabled with force DirectShow decode Disabled

[tebasuna51]

Here is a sample pack with my encodes.
Comes with raw WAV, 5.1 AC3, 5.1 DTS, 5.1 AAC (one increase volume enabled and one disabled)...

The wav source in your sample is 2.0.
To compare you must include the 5.1 wav source from the ac3, dts, aac was encoded.

[Seraphic-]

The wav source in your sample is 2.0.
To compare you must include the 5.1 wav source from the ac3, dts, aac was encoded.

That was the "raw" source, it was an upmix.
Included the 2.0 since thought some might want to do their own upmix.

Here are the six mono wavs though. Download (http://temp.seraphicgate.com/monowavs.rar)

I can only comment on 3. the rest are eac3to options I guess?

Re 3: If your users are using mostly software players there should be no problem, most current hardware players with AAC support don't do more than 2 channels AFAIK.

So for those using hardware players that support two channel AAC, if they were to input at 5.1 aac, would it still play using the front left/right channels or not at all?

Also, does Nero have any "paid" version of their AAC encoder that offer with higher efficiency/quality then their "free" one? http://www.nero.com/eng/down-ndaudio.php