[MindsMirror]

I bought TesseracT's new album Polaris, with the bonus DVD which has a stereo 96/24 and a 5.1 mix in AC3 and DTS 96/24. I feel like sharing my thoughts and discoveries regarding the different versions, so that is the purpose of this thread.
 
 
First of all, the standard CD sounds great. It scored DR7 on the TT DR meter. The DR score is not the greatest, but it sounds more dynamic than the meter would suggest. I would say sonically it is at least as good as One and Altered State, which already sounded great.
 
 
The 2.0 channel 96/24 audio from the DVD sounds just like the CD to me. The DR score is identical to the CD, and the levels are practically the same as well.
Here is the spectrogram of one of the tracks from audacity.
 

 
There's nothing above 22.5KHz. I tried converting the CD to 96KHz and subtracting it to see if they are truly identical. There seems to be a very small EQ difference because almost everything below 1KHz was completely subtracted but above 1KHz it was not. It's not enough for me to hear or tell you which is which. This version is pretty much a pointless waste of HDD/DVD space.
 
 
The surround sound mix is a bit confusing. First of all, it scores DR11, or DR12 if you downmix it first.
 
The most noticeable difference is in the vocals on the center channel. I don't have 5.1 speakers so I don't know how it would sound directly like that. If you mix the center channel into the L and R channels at the standard -3dB, it is too loud. On some tracks it is only slightly louder than the CD mix, on others it is significantly louder, and on Phoenix the center vocals completely overpower everything else. The center channel waveform is not really clipped (it touches 0dB for only one sample at those red lines), but it is obviously limited in many sections in Phoenix like this one. Center channel is the third waveform, by the way.
 

 
I thought maybe the center channel was not meant to be mixed at -3dB, but I used a software called Ac3Tool which can analyse the AC3 stream for the intended center channel mix level and it said that it was -3dB. If you don't include the center channel at all, some tracks sound fine, but then Hexes has significantly quieter vocals compared to the CD, and some vocal parts are completely missing from Dystopia. To get the vocal levels to sound similar to the CD on all tracks, I had to mix it at -6dB on Dystopia and Hexes, and -14dB on the rest of them.
 
With the center vocals fixed, the surround mix does still have significant differences to the CD. Certain tracks are EQed and leveled differently, and you can notice the increased dynamic range on the drums. There is noticeably less compression on the loudest sections of the songs.
 
The spectrogram looks just like the 2 channel 96/24 version with nothing above 22.5KHz, even from the DTS 96/24 stream.
 
 
The CD sounds great so you could certainly be happy without the DVD. I would only listen to the DVD if you are able to modify the gain of the center channel. Maybe there's someone who really loves Dan's vocals and wants to hear more of them, who might like the surround mix as is. Also I guess if you hate Dan's vocals and want Ashe O'Hara back in the band, you could listen to the surround mix and take out the center channel.

 

[Tallis161]

Great article, thanks for doing the analysis and writing it.
Loving the new album, and my personal 2 cents is getting dan back in was an extremely wise career move for the guys.
 

 

[sledgeharvy]

I'm glad I'm not the only one who loves this album.. I feel this is an astounding album (sound quality and engineering). I use this when comparing IEMs and players. All instruments have a specific placement in the sound spectrum, which really lends towards the overall sound of the album. 
Also, I can atest, that even down converting it to 320 Kbps mp3 it still sounds just as clear as the lossless version.Yes, you can tell the difference, but you have to really scrutinize it. Blindly, you wouldn't tell the difference. 

 

[dtfreak05]

I just stumbled upon this thread after ordering the copy of this album with the 5.1 mix, and I have what is probably a stupid question. I've down-mixed several of Opeth's and Steven Wilson's albums from 5.1 to stereo and I've loved the results, so I'm excited to do it with this album, I normally use Adobe Audition (it's a little easier to work with, for me at least, than Audacity) to load the tracks and mix them to stereo. But, I've only learned to do this from another post on here http://www.head-fi.org/t/93206/darkest-desires-part-v/2250, and I'm pretty much a n00b when it comes to audio mixing. So, my question is, to do as you suggest with the center track levels, would I normalize the center track individually to -6db (or -14), and then fold it into the rest of the tracks? Or is there a different process that's necessary? This really is a great album, thanks for taking the time to write this post.

 

[MindsMirror]

I just stumbled upon this thread after ordering the copy of this album with the 5.1 mix, and I have what is probably a stupid question. I've down-mixed several of Opeth's and Steven Wilson's albums from 5.1 to stereo and I've loved the results, so I'm excited to do it with this album, I normally use Adobe Audition (it's a little easier to work with, for me at least, than Audacity) to load the tracks and mix them to stereo. But, I've only learned to do this from another post on here http://www.head-fi.org/t/93206/darkest-desires-part-v/2250, and I'm pretty much a n00b when it comes to audio mixing. So, my question is, to do as you suggest with the center track levels, would I normalize the center track individually to -6db (or -14), and then fold it into the rest of the tracks? Or is there a different process that's necessary? This really is a great album, thanks for taking the time to write this post.

I have come across that thread before, someone asked me about down mixing the Still Life album and referred me to those instructions. To me it makes the process seem unnecessarily complicated. You don't even need to use Audacity or Audition, foobar can do all of those steps very easily. Those instructions do not tell you to mix the center and surround channels at -3dB like you (normally) should. That method using the foobar DTS decoder also does not allow you to decode the lossless DTS 96/24 stream present on the Polaris DVD, so you end up with just the lossy core instead.
 
What I did and would recommend for this album is use AudioMuxer with the ArcSoft DTS decoder to decode the DTS 96/24 audio and convert it to 5.1 channel 96KHz 24 bit WAV. Then use the foobar converter to go the rest of the way.
 
Normally you could just use foobar's built in "downmix channels to stereo" DSP, but since the center channel is off on this album you need to adjust it manually, I would use the matrix mixer component for that. You can do it either way, by applying -3dB or -11dB to the center channel and then mixing it as you normally would, or by manually setting the center channel mix level straight to -6dB or -14dB. The former is probably easier to do. In the DSP chain, add one instance of matrix mixer and set the value of FC/FC to 0.707 or 0.282 (leave the rest of the diagonals at 1), then add an instance of "downmix channels to stereo" which will automatically apply the other -3dBs of attenuation to center and handle the rest of the mixing. Make sure you also reduce the global gain to prevent clipping when mixing the other channels in, you can do that with a global multiplier in matrix mixer.
 
Let me know if you need any more explanation. I know AudioMuxer and the ArcSoft decoder where not very intuitive to set up and use, it took me a while to figure it out.

 

[dtfreak05]

Thanks for your response! I'll try that, though I've never used AudioMuxer so hopefully I can figure it out, my DVD will hopefully be shipped this week so I will report back when I've had a chance to attempt this.

 

[dtfreak05]

Alright, I've received my DVD and I've extracted the 96/24 DTS audio to FLAC and I'm in Foobar trying to figure out the Matrix Mixer. I believe I have the channel levels set right, but the global gain multiplier is throwing me off. No matter what I set it to, I am given a massively loud file when I look at it in Audacity. Should the normalize box be checked? What multiplier dos you use? I had been using the ReplayGain album analysis and applying attenuation with that, if I add that to the chain will that achieve the same result instead of using the multiplier in MM? Screenshot attached. In this screenshot I have the multiplier set to 9 because the ReplayGain album analysis showed that the album gain should be -8.63 dB, but I'm not sure what units the multiplier is actually referring to.

 

[dtfreak05]

And this is what my processing chain currently looks like, with ReplayGain applied. This is the only way I've been able to produce a good looking waveform so far, without an understanding of the multiplier in MM.

 

[MindsMirror]

The global multiplier and all other values in Matrix Mixer are not in decibels, they are linear. So 1.0 would be no change, 0.355 would be about -9dB, zero would be dead silence, and -9 would be the signal flipped completely upside down and nine times louder.
 
Your matrix settings are all good. The normalization in MM won't do anything to that matrix you are using because you aren't using it to mix channels to a level greater than 1.0. If for example you mixed 1.0 FL and 1.0 FR into FC, the normalizer would apply a global multiplier of 0.5 so that FL + FR would not exceed 1.0.
 
You don't need to apply both replaygain and a multiplier, they each do the same thing, you only need to do it one way. It sounds like you've achieved a good enough result with replaygain so that's all you need to do. If your file does not clip and has some room still, you can go back and do it again, but add a dB or two to the replaygain preamp, to get the file as loud as possible before clipping.

 

[dtfreak05]

Thanks! That makes sense. Sounds like it will be some trial and error with the multiplier to get it at a good level. I've liked my results with ReplayGain so far, but it seems to be a bit too conservative with the attenuation as I am left with fairly low volume files. Great sounding, but low volume. So I might do it by track instead of by album. I'd like to get a consistent workflow set up but it may have to be tweaked for each album I suppose. Is there a good rule of thumb for applying a multiplier to the global gain on most albums?

Another question, regarding the channel levels...does Foobar know what levels to mix the individual channels at based on data in the files? I've used Foobar to downmix The Raven That Refused To Sing, and most of the levels are fine, except for a couple spots where a lead instrument seems a bit low in the mix. However, the Deliverance remix from 2015 is pretty much unlistenable due to low volumes in some of the channels, but that's also a Bruce Soord product so maybe some of the same issues are present there as with Polaris. My understanding is that AC3 files contain that data, but I don't know if the DTS streams do. I would think they'd have to for normal home theater playback but I'm not sure how that works.

 

[MindsMirror]

There's not a good way by default to know ahead of time how much gain you need. The replaygain peak level tells you the non mixed level. I have found a really usefull component called Dynamic DSP Mod which, among other things, can tell you the replaygain stats after applying DSP like the mixer, but the site it was on just moved and it looks like the download link is broken. Most 5.1 albums I've seen peak at somewhere from 1.4 to 2.1 after mixing to stereo, so you need about -4 to -7 dB gain.

 

[dtfreak05]

Ah! I did not know that about ReplayGain, though I suppose I should have realized that. Thanks for all your help, I now feel like I have a much better understanding how to do this properly. Do you have any input on the situation with the individual channel levels? Or is that something that just needs to be addressed on an individual album basis? I know you used AC3Tool, I haven't been able to figure that one out yet. Is there a good GUI for that (I'm worthless when it comes to command line stuff)?

 

[MindsMirror]

The individual channel levels should be the same for any 5.1 album. Center is mixed at -3dB because -3dB from two speakers is as loud as 0dB from one, and the surround channels are at -3dB because they are behind you and naturaly sound quieter. Foobar's downmix to stereo component will do that always. The only reason I did something different on this album is because it sounded wrong to me. It would probably sound wrong even if you played it on 5.1 speakers. Ac3tool didnt actualy tell me anything usefull, it just said to mix center and surround channels at -3dB like usual.

I think I also did some adjustment to the center channel in Deliverance. I guess Bruce Scord hasn't quite got the hang of surround mixing yet, or he has different personal preferences to mine.

 

[dtfreak05]

Cool! That's good to know, now I can work on going back and redoing all the ones I did in Audition with some better gain levels. I'll probably still use Audition to load up the individual channels (like on Deliverance) and see what needs to be changed, some of that needs some gain as opposed to attenuation. Thanks again!

 

[MindsMirror]

I think I also did some adjustment to the center channel in Deliverance.

Cool! That's good to know, now I can work on going back and redoing all the ones I did in Audition with some better gain levels. I'll probably still use Audition to load up the individual channels (like on Deliverance) and see what needs to be changed, some of that needs some gain as opposed to attenuation. Thanks again!

I just got home and checked what I did to Deliverance. I actually removed the center channel entirely, which made the level of the vocals sound about the same as they did on his CD mix.