dcadec icon indicating copy to clipboard operation
dcadec copied to clipboard

DTS-MA channel tests

Open madshi opened this issue 9 years ago • 6 comments

tebasuna from doom9 was kind enough to do custom encodes to test all those weird DTS-HD speaker configurations. Here's the download:

https://www.sendspace.com/file/mtejw5

According to his tests, dcadec decodes them all bit perfectly. The one thing I'm not sure about is the channel masks. They might be correct, or maybe not. I haven't really checked. Anyway, I thought you might be interested in the samples, so here they are. Maybe if you find the time, you could double check that the WAV channel masks produced by dtsdec are reasonable.

In any case, you can close this issue, if you like. The main purpose of this issue is just to share the samples.

madshi avatar Apr 04 '15 08:04 madshi

I was completely lost in that post that tebasuna posted. It was overelaborate, I find it difficult to understand his English, and I don't know what the asterisk means that he uses between channels. He was suggesting channel masks that included top speakers. IMO the strange setup surround channels should just mix 1:1 into standard 7.1 surround channels.

Ls -> Side Left Rs -> Side Right Lsr -> Back Left Rsr -> Back Right

I don't need eac3to producing a file with a channel mask of 22031, from a strange 7.1 setup. That's nonsense. There isn't even any height information in the first place from the strange 7.1 setup. 7.1 should always be 63f, 5.1 should always be 3f, etc. I really don't understand where he is coming up with these layouts.

ramicio avatar Apr 06 '15 20:04 ramicio

What people usually refer to as "strange setup" all results in default 7.1 masks with dcadec now. There is basically 3 "normal" 7.1 masks, all of which result in the same WAV mask now.

There are however a long list of other 7.1 setups which use one or more height channels. Personally, I've never seen one used on a Blu-ray, but dcadec should export them using the best approximation in the WAV channel masks available (judging from the speaker mask matching code).

If the codec says "this is a height" channel, then why should we just blatantly ignore that? Maybe it is a height channel, and just playing it through a normal 7.1 set may sound rather weird?

The big discussion was about the three 7.1 layouts without height channels, one of which is the "strange setup" people seem to refer to. Those now all map to the same WAV layout, ordinary 0x63f.

Nevcairiel avatar Apr 06 '15 20:04 Nevcairiel

Because nothing of those masks he listed are even symmetric. I don't even know how these supposed setups with height exist. They don't even exist in the DTS encoder suite. AFAIK DTS-HD is a 2D format.

ramicio avatar Apr 06 '15 20:04 ramicio

The DTS-HD encoder suite offers these configurations just fine. Make sure to check "Display all 7.1 channel layouts" in the preferences.

Here is a list 7.1 height formats it offers to encode in (only the relevant parts, as the fronts are the same in all) Ls, Rs, Lh, Rh (front heights) Ls, Rs, Lhs, Rhs (side-surround heights) Ls, Rs, Cs, Ch (center surround and front-center height) Ls, Rs, Cs, Oh (center surround and overhead speaker)

Nevcairiel avatar Apr 06 '15 20:04 Nevcairiel

Make sure to check "Display all 7.1 channel layouts" in the preferences.

That would explain why I never saw them before...foot in mouth...maybe there needs to be something in eac3to to warn of the exact channel layout and mask being applied...

ramicio avatar Apr 06 '15 20:04 ramicio

I'll take care of this in eac3to properly, in any case. Just wanted to provide foo86 with the samples, because devs always like weird samples.

madshi avatar Apr 06 '15 23:04 madshi