3ll3d00d

10 minutes ago, MikeDuke said:

Has anybody had a chance to do Shang Chi.  It's getting blasted over on AVS.

Thanks

https://beqcatalogue.readthedocs.io/en/latest/aron7awol/61166706/

On 2/28/2021 at 2:28 PM, Archaea said:

Has anyone measured the content of the 1917 movie?  Near the end there is some ominous low frequency bass. Reminds me of the Hunger Games raffle scene.  Can someone measure to see what frequency range that is on 1917?

you can see it here - https://beqcatalogue.readthedocs.io/en/latest/aron7awol/59350372/

the spike is clearly visible on the heatmap view

7 hours ago, maxmercy said:

Has anyone done any experiments on what AC3/DTS compression does to a spectrogram/FFT?

Only at the low end (you lose content at low bit rates)

what does the average refer to in the CEA-2010 graphs? literally the average of all drivers in the site or something else?

Re the holes on the side, translate says they are "acoustically filtered", does it mean some damping material behind the holes? I looked into doing this once but the amount of trial and error (and/or lack of modelling tools) put me off. One thing I do remember though is that any such speaker tended to be pretty deep in order to get enough delay on the side output to affect directivity in a desirable way (iirc). These are v shallow speakers though by the looks of it. Are you using them for this purpose or something else?

you can also measure excursion using an accelerometer and an appropriate pre amp, you can find details of the schematic in the article linked from https://www.diyaudio.com/forums/equipment-and-tools/267734-accelerometer-testing-loudspeaker-drivers.html (or that guy sometimes sells either a pre amp or just the pcb via a thread on diya)

finally wrote a user guide for the app -> https://beqdesigner.readthedocs.io/en/latest/

BBC4 invariably has some interesting music shows on a Friday night, this one popped up tonight - https://www.bbc.co.uk/programmes/m00023xl - which I thought I'd post here given the subject matter, might be of interest (if those outside the UK can see it that is)

https://github.com/3ll3d00d/beqdesigner/releases/tag/0.5.2 contains some enhancements to the built in bass management sim which means you'll get a more accurate idea of whether it is going to clip or not after filtering & also lets you more easily see the impact of BEQ on the full track (via the spectrum comparison view). If anyone has any other suggestions for how to make this more useable then feel free to raise them (otherwise it's just whatever I think makes sense based on me using it to try out BEQ for tracks).

fwiw this release can show the bass managed result of a set of filtered channels in the app, updates live as you change filters and reports on headroom available - https://github.com/3ll3d00d/beqdesigner/releases/tag/0.5.0

I recommend using a decimated track if you want this to perform acceptably so that means it isn't a perfect simulation, close enough though for this purpose IMO

FWIW there was an error in the convert to dB function which put a lower limit of ~ -92dB on the resulting value, the actual response is

https://imgur.com/a/EiDaMN0

29 minutes ago, SME said:

I'm not the least bit surprised that the resulting mono tracks look very similar, even for cases in which the results sound totally different.  In fact, I would expect this to occur especially when people are using your app to create the BEQs visually as opposed to doing it completely blindly and by-ear.

Has anyone created a BEQ by ear alone?

No I am not, and have not been, missing your point. I agree that a per channel solution's should be better and is better in theory. I question whether it is practically that much better in practice on certain tracks (and whether the effort involved in creating the pre channel beq is worth it). I agree it would take a per channel comparison to get a more informed view.

I commented in the first place because of an idle observation (possibly in the avs thread) that the resulting mono tracks don't look that different (presumably because it is dominated by the louder channels). I haven't done any detailed comparison myself though hence why continued discussion in general is a bit pointless

I will dig out the relevant graph to illustrate later and see if it matches my initial idle observation.

10 hours ago, SME said:

I do agree that the post BEQ surround channel averages look odd, particularly below 30 Hz.  Why do all the surround channel curves converge to one curve below there?  I would not expect that to happen.  The shape of the curves and lack of finer details is also unusual.  It looks like it could be garbage.  Maybe insufficient precision somewhere?  Are you applying the 1st order high pass at 10 Hz?  And is it working correctly?

that is strange, I haven't noticed that on any other tracks. It looks like the analysis of the source signal has basically dropped down into the abyss. I'll have to look into that one.

https://imgur.com/a/UoNlJuS

re the rest of the discussion, it seems a bit pointless to discuss further in abstract terms as it hinges on one's definition of subtle vs marked and whether an effect at -x dB is one or the other. 

perhaps this is clearer

before

https://imgur.com/a/lyOBXiQ

after

https://imgur.com/a/n8g8KHe

this is the channel levels on the track so in reality LFE would be another 10dB higher

this is average but the delta between the channels is similar on the peak chart (just much harder to read)

The post beq surround channel looks odd to me (i.e. it is just the filter shape) and it's at a *much* lower level than the LFE and C. Even if there are distinct effects in the surrounds that aren't in any other channel, IMV it's going to be at most a subtle difference.

The surround specific bass boost requires an extremely large boost in order to bring it up to a pretty low level relative to the lfe hence, in this case, even if you have channel specific beq, I suspect you will struggle to notice it. 

On 11/12/2018 at 9:04 PM, SME said:

FWIW, I liked TLJ as a film a lot more after the second watch.  It's definitely flawed, but what SW film isn't?

I just found it a bit dull compared to TFA or Rogue One, storyline seemed particularly  derivative of earlier films as well so that it felt like a remake at times rather than a new film.

On 11/12/2018 at 9:04 PM, SME said:

This is probably why almost every one of the BEQs developed by @maxmercy uses quite different filters on each channel, and why I believe this approach is usually necessary for good BEQ sound quality. 

in principle I agree, in practice I'm not so sure for a track like this where the surrounds are so much lower in level and even LR is another few dB down on the C. I haven't compared but I would not be surprised if there was a pretty small audible difference between the two approaches in this case.

I found TLJ quite underwhelming even with the volume turned up and with BEQ on though perhaps my impression is coloured by my impression of the film (which I also found pretty underwhelming). FWIW I posted the per channel pva for that on avs - https://www.avsforum.com/forum/113-subwoofers-bass-transducers/2995212-bass-eq-filtered-movies.html#post57055584 - as I was curious about the relative merits of the two BEQ approaches (pre and post). It seems to be a really heavily filtered track, almost looks like they baked bass management into the track itself.

bit quiet round here recently but in case anyone is watching... latest beta builds have a bunch of features around seeing the impact of filters on the waveform and being able to zoom into the spectrum for slices of the track while also seeing the overall track. The avs thread has some pics and details - https://www.avsforum.com/forum/113-subwoofers-bass-transducers/2995212-bass-eq-filtered-movies.html#post57032634

latest builds have a split screen before/after beq spectrogram view as well as a way to check the headroom in the waveform pre/post BEQ which you may find useful (and colourful!)

there is a release of that feature available at https://github.com/3ll3d00d/beqdesigner/releases/tag/0.4.1 & described in https://www.avsforum.com/forum/113-subwoofers-bass-transducers/2995212-bass-eq-filtered-movies.html#post56940944

ok problem solved, adding an aformat step (to convert to dbl and then back to s32) around the biquads fixes the issue.

Looking at it some more (and checking the code at https://www.ffmpeg.org/doxygen/2.8/af__biquads_8c_source.html) suggests it really should be possible to avoid this problem unless something else in the chain clips it. Will dig a bit further.

I like the quick test

https://imgur.com/a/8HopqIx

which shows there is zero headroom available. I suppose this is not that surprising as ffmpeg filters are treated as completely independent blocks.

not ideal though perhaps not a blocker as it would be simple enough to attenuate before filtering, good enough for analysis purposes perhaps (though not for the "create remuxed file" case)

sample command for reference - https://gist.github.com/3ll3d00d/40be3ec6e1a5c0466ae324350be65cb0#file-gistfile1-txt

I think I've worked out the incantation required to get ffmpeg to apply filters to individual channels so was thinking about how to test whether it does this correctly (re the concerns mentioned earlier about clipping). Any suggestions on what a simple, easily repeatable, test would be for this? i.e. generate a specific signal, apply some filter, output should be x.