You left out the center channel when you down mixed. You probably just included the front left and right channels.
+ Reply to Thread
Results 31 to 51 of 51
Any number of ways. But we don't have any MediaInfo about these files, nor any samples with which to even make a guess.
Plus, it's likely that what you intend to do cannot be done - PERIOD, or cannot be done without major modification to the file/stream (with LOSS as consequence). I could go into detail at length, but will spare you and myself and others, because frankly, you have yet to be forthcoming.
Again: RE-EVALUATE. Your tools & methods. Your needs & assumptions.
mkvextract isn't lossy. What makes you think it is? You might want to try MKVCleaver. It extracts the same way, but I find it easier to use.
How are you converting the 5.1ch audio to stereo? Which program are you using?
Here's the HD Streams Extractor opening and downmixing the audio in an MKV. I'm using the version built into MeGUI but it works the same as the standalone version.
The audio happens to be AC3 in this case and I've elected to convert it to FLAC. I've typed "-downStereo -normalize" into the option section (without the quotes, so the HD streams extractor will downmix and then increase the volume to maximum.
The HD Extractor will only extract and convert. If you want to convert already extracted audio you could try TAudioConverter, although it should also be able to downmix to stereo and it'll also extract the audio from video files for you so there's no need to do it manually.
Or if you happen to have MeGUI installed you can load all sort of audio files (and some video) into it's audio section, pic an audio encoder and then configure it to downmix to stereo.
When you listen to 5.1ch audio using a stereo system, it's downmixed on playback. There's absolutely no reason why you shouldn't be able to downmix multichannel audio to stereo while encoding and have it sound exactly same. There's minor variations to the way audio can be downmixed, but that aside it should be downmixed the same way whether it's downmixed on playback or it's downmixed when converting.
Here's a few more audio conversion programs. A couple of them will convert the audio in video files. I haven't used any of them myself, but I'm sure they'd all downmix to stereo.
Free Audio Converter
Pazera Free Audio Extractor
TFM Audio Tool
Thank you hello_hello. TAudioCoverter did work for me. I will look through other suggestions as well.
Ok. I am trying to extract music from movies. So I had extracted 5.1 channels of audio and I'm trying to remove SFX from the audio leaving just the music/ little SFX. Currently I am trying to use SpectraLayers Pro to do the job, and here is the question. Should I edit my audio 1 channel at the time or edit them using stereo?
For what I found out is that, if I use stereo, certain frequencies will be cut off as well as some of the music. How should I edit my audio such that no music is omitted so I will have the best of a cleaned audio with no/little SFX?
What is on L channel?
What is on R channel?
What is on C channel?
What is on Ls channel?
What is on Rs channel?
What is the proportion of Dialog to SFX to Music in each of those channels?
Are they steady-state or do they vary greatly in composition during time?
"Spectrally", what are the frequency ranges of the SFXes & Musics (I know what range dialog would be)?
These are the kinds of questions you would need to answer first. And THEN, if you had the patience & skill, you could MAYBE get a moderately, incrementally improved balance, along with moderate quality loss in the remaining audio as a result.
I'm not joking around with you here, either. I have done this professionally many times before. That IT IS NOT EASY is putting it mildly. There are tricks you can do with Channel Math that might help a touch, but I need you to get realistic here.
Why do you need lower SFX? or more Music? There could easily be other, better sources.
Ok. This is something new and I trying out. I've seen people extracting audio and clearing up SFX from movie before, and I am trying how to do that. Honestly, I have no experience in doing this and this might be the first time. I am sorry if my questions might not be specific enough. I was just trying to remove SFX from 5.1 audio.
Is your point to do this just as an exercise? What is reasoning for the need? That may seem like an intrustive and presumptive question, but it has a great deal of bearing on how I (or others) could direct you.
Example: are you doing this because, in YOUR opinion the SFX gets too overpowering at times? Then just try some dynamic range compression, maybe adding on some envelope-following bandpass filtering. That should make the SFX much more balanced with the Dialog, though it would also reduce the Music.
Each element has certain properties or qualities that make it unique and can be exploited, but ALL elements also have some things in common, so modification of one always includes some portion of modification of ALL (usually in a bad way). And there is no magic wand or miracle app, contrary to what is advertised or shown on NCIS. Even Spectralayers, as groundbreaking and interesting as it is, still has to work within normal parameters of the physics of sound. There still isn't (and won't be for a while) any algorithm that can even come close to differentiating objects and patterns as well as the human mind.
I am actually learning how to remove SFX from audio through Sony SpectraLayers Pro. The reason why I am learning this is because I had listened to many complete scores of movies and want to learn how these people remove SFX from the music. As you may know that not all film music is available in stores and soundtracks and sometimes we just need to extract those music from the film itself and clean up film SFX to have a cleaned SXF-free music.
Which people are you referring to?
Sound mixers for films start with everything on separate tracks. Sometimes 24, 48, 64, 96, 100's even. They want to do a submix with only Music, no FX?- They just pull down the FX subgroup master fader.
In fact, movie mixes are usually made into what's known as "stems": Separate submixes with 3 stereo (or multichannel) groupings (Dialog, Music & SFX). So for a stereo master film, you'd use 6 channels (actually 8, because they usually expect a standard stereo mixdown reference set as well). For 5.1 films, there are usually ~26 tracks (6 for each group of the 3, plus 6 for the mixdown, plus 2 for a stereo mixdown). That makes it easy to replace.
IME, it's quite rare that a score of a famous movie is not also available separately.
BTW, there's another trick that is used for soundtracks: repetition fill-in. Since most scores have elements of repetition or cyclicality in them, one can often take a section that is more isolated and use it to cover/replace a similar section that is NOT as isolated.
Remember this analogy: unmixing an element from a sound mix or a video composite is about as difficult, time-consuming and successful a job as removing an ingredient from cake batter after you've mixed it. Theoretically possible (sometimes more than others depending upon the ingredients and completeness of existing mix) but never 100% successful and not worth the exhaustive effort.
Oh, and using lossy compressed sources just compounds the problem.
So, should I work on my editing using mono or multichannel? Since it is already been mixed like you said, it is like removing an ingredient from cake batter after mixing it. For example, the 4 selected channels in the picture when played, there isn't much SFX and vocals where as the 3rd channel is where most SFX and dialog are placed. If I just use the 4/5 tracks (excluding 3rd channel) I might will lose some music as well. So I'm not sure if I should extract the audio to 6 individual mono wave and edit them one by one.
Last edited by wildsharks; 8th May 2014 at 19:21.
There is no RULE. You have to try ALL options, play with it, and find the best one that works for your target outcome. And this changes every title you work with. It's all manual skill & understanding from here on out. Which is why I said it would be difficult (and painstaking and time-intensive and probably expensive...), even for the best of us.
Ok, I know for a 5.1 channel, the Centre is for dialog/music vocals, Front L/R is for Music, Surround L/R is for sound effects and Subwoofer is for low freq. effects. How to tell which is which from the picture above?
Also, why are there echoes of music/dialogue when selecting certain channels?
I don't know if I am doing it right since I am just a starter. What I did was that, I use all the channels and eliminate away the centre channel (since that channel has a lot of vocals and SFX) and converted the other 5 channels into stereo.
I am currently using this software 'SpectraLayers' to clean up my music and at the same time learning how to use it. For what I know, when I tried cleaning, my music becomes kinda squeaky and still leaves some SFX. Not sure how to fix that as well.
So, is there a way to deal with spectral overlaps to identify noise in over laps?
Does anyone know the best way to remove vocal and sfx echos from an audio?
There is no "BEST WAY" when you're already starting from MIXED sources. You will A*L*W*A*Y*S have overlap (also known as crosstalk)†. Just do the best you can and live with the result. There are many tricks, but teaching them amounts to teaching you how to be a perceptive Audio Engineer, and that takes much longer than a simple web tute. Even with the knowledge & the tricks, it is quite imperfect.
The main trick is to start with an UN-MIXED source.
Seriously, you've already been told this a number of times and you're still asking at post #50?!!!!
†Not counting rare & strangely-mixed non-standard soundtracks that hard-pan their elements. Even then, it's unlikely.
Last edited by Cornucopia; 11th Jun 2014 at 01:01.