Dolby Atmos and REW

[thothsong]

hi thothsong, was hoping you might know how to get test sweeps to front high speakers in a std 7.1 setup (AVR is not atmos capable)? Tried all of yours and they only play on LR or SR/SL using VLC configured in bitstream over HDMI. thanks

I'm confused what you mean by "front high" speakers. If it's really just 7.1, no height/ceiling speakers, then you should be able to use normal REW device output, either Java or an ASIO driver, to access full 7.1, no need to use the Atmos files.

 

[potzie]

I'm confused what you mean by "front high" speakers. If it's really just 7.1, no height/ceiling speakers, then you should be able to use normal REW device output, either Java or an ASIO driver, to access full 7.1, no need to use the Atmos files.

AVR is a Integra DTR-70.4 which is 9 channel with no atmos. I use the side surrounds, disable rear surrounds and have options to use either dolby front highs and or front wides. I have selected front highs to make it 7.1. ASIO 5-8 play on the side surrounds as the rears surrounds are disabled and the AVR remaps them back to the sides. Playing those sweeps indicates DD+ as an input and if i choose the DSP mode as dolby 7.1H or DSX, it makes no difference to the output. No sound ever on the front highs. I assume the FH/FW are not discrete channels, so i have run out of ideas. cheers

 

[thothsong]

AVR is a Integra DTR-70.4 which is 9 channel with no atmos. I use the side surrounds, disable rear surrounds and have options to use either dolby front highs and or front wides. I have selected front highs to make it 7.1. ASIO 5-8 play on the side surrounds as the rears surrounds are disabled and the AVR remaps them back to the sides. Playing those sweeps indicates DD+ as an input and if i choose the DSP mode as dolby 7.1H or DSX, it makes no difference to the output. No sound ever on the front highs. I assume the FH/FW are not discrete channels, so i have run out of ideas. cheers

The Atmos files are encoded into a 5.1 downmix, so if your AVR doesn't support Atmos decoding, you'll only get the 5.1 downmix from them. Windows WASAPI doesn't have support for front high channels, so you can't target them using Java or ASIO drivers either.

 

[potzie]

The Atmos files are encoded into a 5.1 downmix, so if your AVR doesn't support Atmos decoding, you'll only get the 5.1 downmix from them. Windows WASAPI doesn't have support for front high channels, so you can't target them using Java or ASIO drivers either.

ok, thanks. I assume the only option is to connect the highs temporarily to the rear suround output and hope the 7-8 channels work correctly. Just wondered how people would have calibrated highs and wides pre atmos?

 

[thothsong]

Just wondered how people would have calibrated highs and wides pre atmos?

By temporarily rerouting in their AVR, I would guess.

 

[buffkung]

By temporarily rerouting in their AVR, I would guess.

Hello, I have 9.1.8 setup. Is it possible to generate front high sweep file?
Thank you.

 

[thothsong]

I have 9.1.8 setup. Is it possible to generate front high sweep file?

The AWS service I used only supports 9.1.6 Channel-Based Immersive content, sorry.

 

[canavan]

Hello, I have 9.1.8 setup. Is it possible to generate front high sweep file?
Thank you.

Given the number of channels you're using, I'd suspect that you have a AVP that can route external signals to arbitrary channels. If that's not an option, you'll have to find someone with access to a proper Atmos encoder, a DAW and enough knowledge to use them properly. On top of that, encoding such a REW sweep would require figuring out how to snap the signal to the specific channels by placing the objects in the correct position and enabling the speaker snap bit, so it's probably not straight forward, but will require some experimentation.

 

[serko70]

Hello, I have 9.1.8 setup. Is it possible to generate front high sweep file?
Thank you.

If @thothsong can't generate them, trust me noone else can

What gear are you using to decode/feed a 9.1.8 set up if you don't mind me asking?

 

[thothsong]

On top of that, encoding such a REW sweep would require figuring out how to snap the signal to the specific channels by placing the objects in the correct position and enabling the speaker snap bit

The snap bit is an interesting question; Dolby's 5.1.2/7.1.4/9.1.6 DD+ Atmos pink noise files don't actually use it, the dynamic objects are simply placed at the canonical 3D coordinates for the corresponding speakers. I'll also note that Dolby doesn't seem to have a 9.1.8 sell sheet, and their 11.1.8 sell sheet shows the heights layout isn't just x.x.6 plus two more speakers, rather the Top Middle/Side speakers are shifted away from middle. And to further confuse, what's publicly available of the TrueHD Atmos specification has a different layout for beyond-6 height channels, with Top Front Center and Top Center. The E-AC3 JOC specification only lists canonical coordinates for 9.1.6.

 

[canavan]

If @thothsong can't generate them, trust me noone else can

What gear are you using to decode/feed a 9.1.8 set up if you don't mind me asking?

I'm pretty sure thotsong could do it, if someone sent him a few hundred bucks for a DAW and the encoder - and maybe a processor and a few extra speakers to actually test the result.

I intended to contact the starter of the thread over at AVSforums where he published a few cool Atmos object demos, whether he could encode a few test files for me, but so far, the AWS encoder was good enough for me.

 

[thothsong]

At one point I toyed with the idea of buying a 1-year license to the Dolby Media Encoder, because TrueHD Atmos sweeps would be nice to have, but it seems to only be available to companies, not individuals..

 

[serko70]

I'm pretty sure thotsong could do it, if someone sent him a few hundred bucks for a DAW and the encoder - and maybe a processor and a few extra speakers to actually test the result.

I intended to contact the starter of the thread over at AVSforums where he published a few cool Atmos object demos, whether he could encode a few test files for me, but so far, the AWS encoder was good enough for me.

I don't remeber which video (I watch lots) but I remember one of the most sought after audio engineers producing select blockbuster Atmos mixes, admiitting to "an Atmos master mix never to have more than 7.1.2 (or sometimes 5.1.4) channels and the rest just being dynamically produced during playback by the proprieatry algo. I also hear that Sony, Apple and others are coming with competitive immerisive systems (and from a pair of earphones) which will be open source and Atmos's days are numbered if they continue exclusive, expensive and studio only access to their 3D object placement.

I don't understand the idea really. They make measuring and correcting Atmos channels even for enthusiats like us, nearly impossible. Before @thothsong came up with the sweeps, I used to switch banana plugs behind the receiever to at least get the frequency response. And it took me nearly a year after that to finally calibrate Atmos speakers correctly to hear the desired effects of Atmos. Before that I used to wonder what they all hype is about. I am sure 90% of home theatre users are not hearing "remotely" what Atmos is capable of . So people cannot hear its magic and don't see any reason to prefer it. The new generation don't see any reason to listen to surround sound let alone immersive audio. Spatial placement of sounds cannot be uninteresting for any living creature on Earth but it requires precise calibration to get there which the vendor relentlessly deems from very hard to impossible.

Btw @thothsong , I produced many variations of the Atmos sweeps in AWS recently and this set came out 4.5dB louder than the others with one of the settings. Certain Denon/Marantz receivers with odd dialnorm settings will play the sweeps at -12dB regardless of the loudness management setting:

Defeating dialnorm

Dialnorm is a type of metadata typically included in lossy/streaming Dolby Digital (DD/AC3) and Dolby Digital + (DD+/EAC3) tracks which instructs the processor to reduce master volume by an amount equal to the difference between -31dB and the value stored in the metadata. Many of us would...

www.avsforum.com

So, I guess these could be useful at least for that gang. And I still see tiny differences in the midrange of surrounds but I will not open that subject

Optimal Atmos Sweeps 256k - Google Drive

drive.google.com

 

[serko70]

I'm pretty sure thotsong could do it, if someone sent him a few hundred bucks for a DAW and the encoder - and maybe a processor and a few extra speakers to actually test the result.

I intended to contact the starter of the thread over at AVSforums where he published a few cool Atmos object demos, whether he could encode a few test files for me, but so far, the AWS encoder was good enough for me.

You didn't say what your decoder set up is like. The reason I am asking is, I recently came up with a method to calibrate Auro 3D speakers (or Mac users for that matter who could never play the sweeps) for example using Audyssey's own measurements and the results are just as good as you would measure them manually in REW except that you cannot with Auro 3D. So if your spakers .8 ceiling speakers are being driven somehow, there can be a way to calibrate them without measuring.

 

[thothsong]

I produced many variations of the Atmos sweeps in AWS recently and this set came out 4.5dB louder than the others with one of the settings

The metadata differences that I see are:
1. DRC profiles specified (Film Light for dynrng and Film Standard for compr)
2. truepk and loudstrm3s values specified, both of which are loudness measures computed without any gain adjustments due to dialogue normalization and DRC being applied

 

[serko70]

SO, in your experience, is there any problem to use them as measurements sweeps with dimmed volume receievers and offset them -4.5dB if used with normal receivers?

 

[thothsong]

SO, in your experience, is there any problem to use them as measurements sweeps with dimmed volume receievers and offset them -4.5dB if used with normal receivers?

Seems OK to me.

 

[canavan]

At one point I toyed with the idea of buying a 1-year license to the Dolby Media Encoder, because TrueHD Atmos sweeps would be nice to have, but it seems to only be available to companies, not individuals..

I believe Dolby offers evaluation licenses, valid for 30 days or so if I'm not mistaken. I think not being / having an official company should not be a problem. I've created a Dolby account for that, but getting a demo license for a supported DAW was way more of a hassle than just using the AWS media encoder.

 

[canavan]

You didn't say what your decoder set up is like. The reason I am asking is, I recently came up with a method to calibrate Auro 3D speakers (or Mac users for that matter who could never play the sweeps) for example using Audyssey's own measurements and the results are just as good as you would measure them manually in REW except that you cannot with Auro 3D. So if your spakers .8 ceiling speakers are being driven somehow, there can be a way to calibrate them without measuring.

There may be a perfect solution for Auro 3D with REW: contact Wilfried Van Baelen, and ask him if they would kindly encode a sweep for Auro 3D. He was interested when I inquired about encoding sweeps for the Smyth Realiser A16, but that lost momentum, and since my Realiser got lost, I'm not too motivated to follow up.

Edit: I believe you were replying to the wrong post. My AVP can route test signals to arbitrary speakers, and I only run 7.1.4 at the moment, not .8.

 

[canavan]

I don't remeber which video (I watch lots) but I remember one of the most sought after audio engineers producing select blockbuster Atmos mixes, admiitting to "an Atmos master mix never to have more than 7.1.2 (or sometimes 5.1.4) channels and the rest just being dynamically produced during playback by the proprieatry algo. I also hear that Sony, Apple and others are coming with competitive immersive systems (and from a pair of earphones) which will be open source and Atmos's days are numbered if they continue exclusive, expensive and studio only access to their 3D object placement.

The objects are a feature, not an issue for regular playback. At least in theory, they allow the sound sources to be rendered as well as the speakers that are actually available allow (including deviation from the speakers' ideal positions).

I don't understand the idea really. They make measuring and correcting Atmos channels even for enthusiats like us, nearly impossible. Before @thothsong came up with the sweeps, I used to switch banana plugs behind the receiever to at least get the frequency response.

That one is on the manufacturer of your receiver. HDMI 2.1 supports up to 32 channels of PCM, but as far as I know, no sink implements more than 8 so far. At least with Trinnov, Storm Audio one can route those 8 channels to arbitrary speakers. The Smyth Realiser has a 16 channel USB interface that would work about as well as the encoded sweeps.

 

[thothsong]

HDMI 2.1 supports up to 32 channels of PCM

The CTA Short Audio Descriptor for L-PCM, even in ANSI/CTA-861-H, still only has a 3-bit field to encode max channels - 1. So how does a sink advertise support for more than 8 channels in its EDID?

 

[canavan]

The CTA Short Audio Descriptor for L-PCM, even in ANSI/CTA-861-H, still only has a 3-bit field to encode max channels - 1. So how does a sink advertise support for more than 8 channels in its EDID?

Good question. I've got no idea. It's actually HDMI 2.0 that seems to have introduced this, and here is a document for an Altera FPGA core that seems to support 32 channels, and one from Xilinx as well.

 

[thothsong]

It's actually HDMI 2.0 that seems to have introduced this

Oh, right, it's actually a separate audio format, called "L-PCM 3D Audio", supporting 9 to 32 channels. I had forgotten about it. I wonder if any sound system actually supports it (mine doesn't), or any source device for that matter (I don't think Windows does).

 

[SimpleDude]

If you've previously downloaded sweeps, you should download new ones. Here are new 1M sweeps:
[REMOVED, CANNOT POST LINKS YET]
and here are new 256k sweeps:
[REMOVED, CANNOT POST LINKS YET]

With the previous versions, I foolishly assumed the auto-generated Dolby dialog normalization metadata value would be the same for all files, since the content was the same, just on different channels. Bad assumption; it turns out dialnorm varied across the files, so absolute level for the sweeps was not consistent. I figured out a way to make them all the same, hence the new versions.

Thanks for these, had a question regarding volume setting on the AVR when running these, I have a Denon X3800H, normally I set the AVR to 0MV and use -30dBFS sweeps in REW, would I need to change these for playing back the latest 256k sweeps for the atmos channels ?

Also the microphone file to use?, I have a UMIK1 and point it directly up at the ceiling for doing a Dirac Live run, when running these sweeps files do I use the 0 degree file or the 90 degree one for the UMIK1 ?

 

[thothsong]

normally I set the AVR to 0MV and use -30dBFS sweeps in REW

I think you're better off leaving REW at -12 dBFS, and lowering your AVR MVL instead.

would I need to change these for playing back the latest 256k sweeps for the atmos channels ?

The Atmos sweeps should play 13.5 dB louder than REW sweeps at -30 dBFS, at a given MVL setting. If you change REW to be at -12 dBFS, then the Atmos sweeps should play 4.5 dB quieter at a given MVL setting.