Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[mwillems]

I have a question about #24.  When I downmix to stereo using JRSS, I seem to still hear the LFE channel effects.  I tried last night changing my output format back and forth between "2 channel in a 5.1 channel wrapper" (which should down mix to stereo) and "2.1."  In both cases I heard the LFE channel, just in one case it was coming from my sub, in the other case from my mains (to the extent they were capable of reproducing it). 

Can one of the devs confirm whether the LFE channel is actually discarded during stereo downmix?

[Hendrik]

Can one of the devs confirm whether the LFE channel is actually discarded during stereo downmix?

I remember Matt answering this same question once, saying that its being used for the downmix.

[mwillems]

I remember Matt answering this same question once, saying that its being used for the downmix.

It's certainly been my subjective impression that the LFE is being used for the downmix (and keeping it in the mix is my preferred alternative).  I'd just seen several people contend that JRSS eliminated the LFE channel, so I wanted to confirm one way or the other.

[Matt]

JRSS uses all channels, LFE included, when it downmixes.

And of course it accounts for the +10dB standard calibration.

JRSS downmixing preserves the energy balance you would hear on a calibrated system.  There's no magic processing like delays, reverbs, or other gimmicks.  The only gray-area is how "hot" to run the mix (the tradeoff between possible clipping and too low of volume).  We try to strike a reasonable balance, and using Internal Volume will remove any worries of clipping during "perfect storm" mixing conditions.

Also, if you have a low-end system, you can use Options > Video > Adaptive Volume to do real-time dynamic range compression and to give the center channel a boost.

[6233638]

I have a question about #24.  When I downmix to stereo using JRSS, I seem to still hear the LFE channel effects.  I tried last night changing my output format back and forth between "2 channel in a 5.1 channel wrapper" (which should down mix to stereo) and "2.1."  In both cases I heard the LFE channel, just in one case it was coming from my sub, in the other case from my mains (to the extent they were capable of reproducing it).  

Can one of the devs confirm whether the LFE channel is actually discarded during stereo downmix?

To be clear, I did not mean to imply that JRSS discarded LFE - it's just that a standard downmix will discard LFE, and without being able to see what is being used inside JRiver, the "safe" assumption is that it is being discarded.

It would still be useful to see what the matrix is, as the results seem to differ from my LAV Audio downmix, but without being able to record the output from JRiver's Audio Path (with LAV Audio, I can dump it with GraphStudio Next) I don't have a way to compare.

I remember Matt answering this same question once, saying that its being used for the downmix.

Thanks - I did a lot of searching before to try and see what the matrix was for JRSS downmixing, but it never turned up anything. However, being able to narrow it down to one poster:

Yes, it does (although it's actually +10 dB).

This is optional on DSP Studio > Output Format > Source is correctly mastered for +10 dB calibration.  Just leave it checked (like the default).

There is no option for this now, but I assume it is now automatic. Confirmation would be good though.


EDIT: Well there you go. Thanks Matt, that's great to hear.

[mwillems]

To be clear, I did not mean to imply that JRSS discarded LFE - it's just that a standard downmix will discard LFE, and without being able to see what is being used inside JRiver, the "safe" assumption is that it is being discarded.

It would still be useful to see what the matrix is, as the results seem to differ from my LAV Audio downmix, but without being able to record the output from JRiver's Audio Path (with LAV Audio, I can dump it with GraphStudio Next) I don't have a way to compare.

Thanks - I did a lot of searching before to try and see what the matrix was for JRSS downmixing, but it never turned up anything. However, being able to narrow it down to one poster:
There is no option for this now, but I assume it is now automatic. Confirmation would be good though.


EDIT: Well there you go. Thanks Matt, that's great to hear.

That also accounts for the observation you made in another thread (http://yabb.jriver.com/interact/index.php?topic=79218.0) that video that begins in stereo tends to sound louder than video that begins in 5.1 and is downmixed to stereo:  in order to account for the 10 dB calibration difference without clipping, the downmix would have to decrease the baseline volume level of the whole ball of wax by ten dB so you'd get a significant difference in volume.

[Matt]

That also accounts for the observation you made in another thread (http://yabb.jriver.com/interact/index.php?topic=79218.0) that video that begins in stereo tends to sound louder than video that begins in 5.1 and is downmixed to stereo:  in order to account for the 10 dB calibration difference without clipping, the downmix would have to decrease the baseline volume level of the whole ball of wax by ten dB so you'd get a significant difference in volume.

That's true.

In practice, you rarely need to cut the full 10dB.  We revised our mixing philosophy a few years back from "it will never, ever clip (mathematically impossible) but the volume is super low" to a more reasonable compromise for levels.  I talk a little more about this in the post above (I added it with an edit).

If the level is still too low, you can use 'Adaptive Volume'.

[mwillems]

That's true.

In practice, you rarely need to cut the full 10dB.  We revised our mixing philosophy a few years back from "it will never, ever clip (mathematically impossible) but the volume is super low" to a more reasonable compromise for levels.  I talk a little more about this in the post above (I added it with an edit).

If the level is still too low, you can use 'Adaptive Volume'.

So if I understand you correctly, the downward adjustment is not fixed (i.e. is dynamic based on the source material)?  I use internal volume with more than 10 dB of attenuation so I'm not concerned about clipping, but how does that dynamic adjustment interact with system calibration? 

I set my reference level to match the correct calibration point, but if the downward adjustment isn't fixed, it sounds like that reference level will be slightly off by a different amount with every movie.  Not that a few dB matters that much, I'm just curious about the mechanics.

[Matt]

So if I understand you correctly, the downward adjustment is not fixed (i.e. is dynamic based on the source material)?  I use internal volume with more than 10 dB of attenuation so I'm not concerned about clipping, but how does that dynamic adjustment interact with system calibration?  

It's dynamic based on the number of channels in the mix.  But mixing 7.1 to 2.0 will always use the same mixing levels.

I set my reference level to match the correct calibration point, but if the downward adjustment isn't fixed, it sounds like that reference level will be slightly off by a different amount with every movie.  Not that a few dB matters that much, I'm just curious about the mechanics.

Downmixing is going to change Reference Volume.  In other words, 5.1 as 2.0 will play at a different level than 2.0.

If this is a problem, you could calibrate volume using a 5.1 or 7.1 signal (Tools > Advanced Tools > Audio Calibration) so it runs through the downmixer.

Then your 2.0 content will be loud relative to Reference Volume, but in my experience Reference Volume is sort of meaningless with 2.0 content anyway (it's so loud it's painful with most CDs).

[6233638]

In practice, you rarely need to cut the full 10dB.  We revised our mixing philosophy a few years back from "it will never, ever clip (mathematically impossible) but the volume is super low" to a more reasonable compromise for levels.  I talk a little more about this in the post above (I added it with an edit).

If the level is still too low, you can use 'Adaptive Volume'.

Well that explains the difference in volume that I heard between LAV Audio and JRSS - after hearing that JRSS does include LFE, I was thinking that my downmix must have been wrong. Having a reduced volume level from downmixing is not a concern for me, because stereo audio tracks are only played at roughly 10/100 on my system, so even a quiet 7.1 downmix is only up at 30/100 at most.

Would it be possible to have an option that normalizes volume across all videos? Personally, I would prefer that the downmix never has any chance of clipping, which creates a big difference in volume between 2.0/5.1/7.1 tracks.

The reduced volume of 7.1 is not a problem, but the difference between them is. I would like to have 5.1 and 2.0 tracks reduced to be at an equal level to 7.1 so that all video playback is consistent.
An example of where it's a problem is playing a DVD with a 5.1 track (up at maybe 20/100) which then goes back to the menus after the credits and is now playing a stereo track at twice the volume I would normally have it at. (10/100)

[mwillems]

It's dynamic based on the number of channels in the mix.  But mixing 7.1 to 2.0 will always use the same mixing levels.

Downmixing is going to change Reference Volume.  In other words, 5.1 as 2.0 will play at a different level than 2.0.

If this is a problem, you could calibrate volume using a 5.1 or 7.1 signal (Tools > Advanced Tools > Audio Calibration) so it runs through the downmixer.

Then your 2.0 content will be loud relative to Reference Volume, but in my experience Reference Volume is sort of meaningless with 2.0 content anyway (it's so loud it's painful with most CDs).

If it modifies the volume consistently for a given channel set that's easy to workaround with zones, I was just imagining JRSS was doing a film by film analysis (which is not the case).  If all 7.1s are treated the same, there's an easy fix: I can just set up different zones for different channel formats with the reference levels set appropriately.  Then I can use zoneswitch to set it to the correct zone based on the number of channels.  I watch a lot of (older) movies that have stereo (or even mono) soundtracks so getting the references set right for different channel configurations is of interest to me. [EDIT: but that probably wouldn't address 6233638's concern about the difference in levels between DVD menus and the films on the DVDs]

I agree that for CDs (even with replaygain) the reference is pretty damned loud, I just wanted to get my levels right for the various film channel sets.  Thanks for the detailed explanation!

[Matt]

Would it be possible to have an option that normalizes volume across all videos? Personally, I would prefer that the downmix never has any chance of clipping, which creates a big difference in volume between 2.0/5.1/7.1 tracks.

I understand the problem, but I'm not sure how to solve it nicely.

Would you like your 2.0 music played at a reduced volume, so that it more closely matches the level when you play a 7.1 movie?  Or is this 2.0 video only?

What happens if you, hypothetically, play an 11.1 file?

[6233638]

I understand the problem, but I'm not sure how to solve it nicely.

I agree, there's no great solution for it. Is ReplayGain a possibility for video, or is that too complex? And you could realistically only do that with video files, and not discs/ISOs.

Would you like your 2.0 music played at a reduced volume, so that it more closely matches the level when you play a 7.1 movie?  Or is this 2.0 video only?

Well, it seems like this would be relatively easy to do across video, by reducing the volume of the 2.0/5.1 tracks. Even when compared to stereo music, stereo videos are quite a bit louder, because they don't have ReplayGain applied.

The same thing could certainly be applied to music in conjunction with ReplayGain so that all audio passing through MC would be at a constant level. Perhaps having a two-level option would be best? "Normalize audio across video" and "normalize audio across mediacenter" ?

What happens if you, hypothetically, play an 11.1 file?

Silence? It's certainly a possibility a few years from now, but in the near future, I don't see us moving beyond 7.1, and I don't think that reducing volume levels to that of a normalized 7.1 downmix is going to push you below 16-bit.