Like the title say's, i'm trying to convert DTS to AAC and also downmix it to 2.0. Here are the settings i'm using
I can't use normalize peaks, because it will just hang and do nothing and the few time's i got it to do something it would just make a 3mb file with no sound. The setting's i'm using in the pic are pretty much the only setting's that will work as if i use anything else i getting a lot of popping in loud scenes.
So, am i doing something wrong to make it come out like this or is that just how.
Oh, i also noticed that the audio sounds better before i mux it with the video, why is that? I though it's not suppose to do anything to the sound?
If you wanna hear a sample to hear what i mean, i'll upload one later when i get back.
Any help on the matter would be great though, thanks in advance!
+ Reply to Thread
Results 1 to 15 of 15
Last edited by kkiller23; 7th Jul 2012 at 13:21.
Is there a really and truly compelling reason why you must use AAC? Are you married to some kind of Apple technology that has painted you into a corner and made this your only option?
If you're not using AAC because some playback device only understands that, why not just convert the DTS file to AC3? We've got tools like eac3to that can easily do this kind of conversion for you and I think you'd getting better results from AC3. Also, we're not real big on any of Nero's tools and to be blunt it's quite possible that Nero's encoder is half-assed. Some other conversion tool like Handbrake or ffmpeg could very likely give significantly better results if you really must have 2 channel AAC as your output.
First detail: DirectShow decoder <= if it's AC3Filter 1.63, be careful, that version of AC3Filter has a bug which causes clipping.
Also, check if the downimixing includes the LFE channel, if it does, generally that's a bad idea.
Third, maybe Q=0.67 is too low for your DTS audio source.
Last but not least,
I tried AC3 and it just crashes. How do i check the version megui is using? I'm no pro at this stuff, that's why i'm using megui. What would you suggest me to use if megui is so bad?
also here is what mediainfo shows for the audio AAC audio
ID : 2
Format : AAC
Format/Info : Advanced Audio Codec
Format profile : LC
Codec ID : 40
Duration : 1h 56mn
Bit rate mode : Variable
Bit rate : 253 Kbps
Maximum bit rate : 329 Kbps
Channel(s) : 2 channels
Channel positions : Front: L R
Sampling rate : 48.0 KHz
Compression mode : Lossy
Stream size : 211 MiB (13%)
Language : English
Encoded date : UTC 2012-07-07 20:44:06
Tagged date : UTC 2012-07-07 20:44:26
There i uploaded a small sample of a loud scene, you can see how it gets muffled and stuff from that.
Last edited by kkiller23; 7th Jul 2012 at 17:15.
If I'm encoding audio, I always encode using Nero AAC with a quality setting of 0.50 (which is the default). AAC is pretty much universally supported by media players. Not that I think I've seen a lot of devices which don't support AC3, but I think I met a TV with a built in media player not so long ago. It was fine with AAC.
There's nothing wrong with MeGUI. Currently it's updated regularly enough to be annoying at times. I guess it went through a stagnant period a few years ago, but what software hasn't? MeGUI includes a GUI for running eac3to. It's called HD Streams Extractor and it's under the Tools menu.
I can't tell you why you're having audio problems, but I'd use the other decoders in preference to DirectShow to see if you luck improves. If you have something like ffdshow installed it'll probably be decoding. If so is it's volume filter (or anyother filter) enabled.
If you want to try another GUI for encoding the audio, ffcoder comes to mind. I quite like it actually, it kind of feels like a more compact version of MeGUI in some ways. If you set the video encoder to "disable" you should be able to encode just the audio. Once again, ffcoder use NeroAAC to do the AAC encoding.
You can see which version of the various tools MeGUI uses (or most of them) by using the Options/Update menu. Normally when the update window opens it probably only displays the tools for which there are updates. Checking "show all files" will fix that, and the current version number of each should be displayed.
I've got an AAC encoded version of that movie (well the audio part) and I definitely would have remembered if it sounded like that. It's as though the audio level has been raised by a ridiculous amount and it's distorting..... well it is distorted.
Is it just that particular audio or is it encoding audio using MeGUI in general which is a disaster? Maybe it's a DirectShow decoder problem. If you have no luck with ffcoder there's a few convert anything to anything programs out there which are easy to use and free. AnyVideoConverter, FormatFactory..... well those two are a good place to start.
No expert at Audio encoding, but OP complains about "Loud" passages - is it possible the Original is too loud and the encoding process just does the best it can?
To prove if the Nero Encoder is duff.
..You could record 1,pause 2, pause 3 ---and set them e.g. at 6db increments, (or "say" what levels you are using? ie adjusting after the recording is over).
After encoding you'd get a good idea of what levels may be OK.
Hope that helps...
What kind of DTS is it....DVD or Theater? I know that the Theater version would normally sound like that because it's the most bang for the buck when it comes to a cinema presentation....I'm no expert either, just guessing.
What is the audio source? Is it from DVD, a download from somewhere?
The audio source is from a bluray i have no clue what kind it is. I used ffdshow to normalize it a bit and it seems to have helped A LOT. I didn't know i could change settings in ffdshow etc.
It might have fixed the problem, i'm not sure yet. I'm encoding another movie now. I'll see how it sounds after it's done.
If you decode the audio via DirectShow..... well there's nothing wrong with doing it that way as such.... however there's always a chance the DirectShow decoder may be altering the audio in some way as it's decoding, and in the case of ffdshow it has all sorts of filters which can do anything from mixing multichannel audio down to stereo to adjusting the volume to adding delays etc. I sometimes use the DirectShow method of decoding when all else fails or I want to process the audio in some way while encoding, but of course it's fairly important to uncheck all of ffdshow's filters when encoding unless you know you want to use one as you probably won't be encoding it "as-is" with any of them enabled. The ffdshow audio decoder configuration shortcut should be in the start menu and it's a good idea to open it and uncheck all the filters before you start encoding but only do it while ffdshow is not currently running otherwise it mightn't keep the setting (if it's running with an icon near the clock whichever settings it's using when the last instance is closed are the settings it'll remember). Alternatively you can right click on the ffdshow audio icon to ensure none of the filters are ticked.
The "need to remember" is probably a good reason to use one of the other decoders when setting up MeGUI's audio encoding in preference to DirectShow in the majority of cases. Decoding via DirectShow is also possibly the reason why you were having problems converting to AC3. It probably depends on the encoder being used and I'm not sure how it all works in any real detail, but if ffdshow's audio mixer was enabled and it's already downmixing multichannel audio to stereo and you then try to encode using MeGUI while telling it to do the same thing.... it could be a case of the encoder expecting multichannel audio as the source but ffdshow sends it stereo, and it simply doesn't know how to cope.
PS. ffdshow also has a volume "normalizing" feature which can be found in the volume filter. It's an on-the fly- type of normalizing which slowly increases the volume until the peaks hit 0db (or maximum allowed volume increase), instantly drops it again when they exceed it, then slowly starts turning it up again etc etc etc. You definitely don't want to be encoding anything with ffdshow's volume normalizing enabled. Under DirectShow Control ffdshow also has an option to enable dynamic range compression the same as MeGUI. It's different to normalizing and I think only uses the volume information contained within some AC3 audio, but you don't want to encode with it enabled either.
If it makes you feel any better I've done it lots of times.... started encoding while forgetting to check which filters are enabled. I encode video via DirectShow sometimes and the same applies. I've forgotten to check the video filters first many times.
Last edited by hello_hello; 8th Jul 2012 at 23:05.
Hello guys! Sorry if i'm making some off topic but i believe my problem is very close in it's basics. I'm seeking for a person that is competent enough to help me with creating solution for audio normalization while encoding from DTS/AC3 5.1 (inside MKV or m2ts) to AAC LC 2.0 (mp4) for the aims of HDS video on demand streaming. For content preparation we use open source: FFmpeg, H.264 and some additional tools - all under Debian/Centos. We have already tried several solutions to apply Replay Gain for such AAC encoding incl AAC Gain but they are not working as good as we want, actually we have troubles with clipping and some other issues - i can explain and show all in details.
Important: I'm ready to pay a very good money for that help (payment in advance acceptable) I don't have strict time limits! I have a team of streaming experts but that problem with sound will soon cause insomnia! If somebody is interested please contact me here or using PM system.
What's you're definition of "normalizing"?
If you're referring to increasing the volume until the peaks hit 0db then many programs capable of converting audio will do it.
If you're referring to adjusting the volume so all sound about the same level, then ReplayGain is probably your best option. It's not (in my opinion) perfect for audio soundtracks (for standard music tracks it's great) but it should get you pretty close.
Depending on the source, it might be better not to normalize at all. For instance if I'm converting a bunch of TV episodes from DVD, then the audio will already be "normalized" in respect to speech being the same level, so I never touch it. Unless I'm downmixing multichannel audio to stereo, in which case I convert it while mixing down to stereo and applying a 6db volume decrease during the process. The reason for that is because simply "combining" multichannel audio can result in peaks above 0db, and after a lot of experimenting I've concluded that for DVD/Bluray audio, a 6db reduction is the most you'll require 99.99% of the time.
Most programs use 89db as the target ReplayGain volume and won't let you change it. It's the standard volume used when applying ReplayGain to music tracks, however it'll very likely cause clipping when using 89db on video soundtracks. I think ideally you want the target volume to be 82db or 83db. When I say clipping..... there should be no clipping when compressing the audio as compressed audio can store values above 0db, however the audio may be clipped on playback. Can you actually hear clipping?
Anyway.... if you want to use ReplayGain and also prevent clipping, assuming the ReplayGain target volume used by your converter is fixed at 89db, apply ReplayGain and then apply a further 6db reduction and you should be safe (well I guess it'll depend on where the audio comes from). As long as the whole process stays in the digital domain you shouldn't cause "clipping" during the process. I use foobar2000 (it's an audio player) for much of my converting. Using foobar2000 you can set up conversion presets and it's pretty easy to set them up to apply ReplayGain (the audio track must contain the ReplayGain info or you'll need to scan it first), then also apply a further gain reduction while converting. It also has a 5.1ch to stereo plugin.
Thanks great for your reply and comprehensive description of your approach. I can answer that i support the second definition of "normalizing" and yes - i actually hear clipping. I can say that your approach regarding downmixing 5.1 to stereo sounds cool but our experience display that after downmixing we often have sound volume decreased and sometimes very seriously! That's the reason we apply amplification for stereo after that to normalize the volume in compare with source. I also can mention that foobar2000 wich i use for working with FLAC on my PC for personal use is not suitable for that project cause it's not a soft for automatized server-based encoding on NIX systems. If you or somebody here is interested to get acquainted with our project and start working on that issue your still welcome and all business details can be discussed.
Well I can only guess as to why your volume is decreasing after downmixing multichannel audio to stereo..... you don't have to think about it too hard to realize when you combine audio, it gets louder. Maybe you need to downmix with something else.
I think in the case of foobar2000 it simply mixes the audio together, while applying a 3db reduction to the center channel (so the centre channel doesn't get louder) and the same for the LFE channel.
If you're hearing clipping..... well I've found audio tracks with peaks of +7.5.db, maybe even a little higher sometimes, but my hardware seems to have the headroom to deal with it. I can't hear any actual clipping.
You can check yourself - just follow this link http://22.214.171.124:8080/index.html#play and enter in the field Stream name: /vod/avatar_1080p.mp4 - after that click Play. Start listening from 01:27:00 for sample with Headphones preferable and with adequate sound volume (not quite) and you will hear!
Yep. I can hear it.