The Low Frequency Content Thread (films, games, music, etc)

[Nicke]

I love the ring trilogy and also the sound/mixing of it,however i have not heard anything that Really stands out in them basswise (kind of darla tapping tank,the .50 cal scene,door knockings in haunting and so on) it just have an amazing soundtrack without any glory moments,thats fine me.

[maxmercy]

I don't understand your graphs. 

 

My experience with this disc is that the low end is not at -28dB with single digits -10dB below that.

 

Without any sort of calibration method, using comparison to other discs and the same system playback parameters, I just don't conclude that FOTR is a green/yellow soundtrack.

-28dB is in reference to the maximum signal possible: all 8 channels with a time coherent 0dBFS signal, adding up to 128dB. So -28dB is still 100dB at the listening position at that single frequency. I use a 2 second FFT in order to have 1/2Hz resolution while not losing transients as much as higher resolutions, and use zero offset. Using offset makes everything look hotter than it actually is, and the higher res the FFT, the larger the time smear.....it's a preference.

 

Our graphs match in relative level for each freq. I use a different maximum so that I never clip, making things look lower in level, but the PvA shows the real deal for each scene... Compared to WotW, TIH and HTTYD, FOTR has less <20Hz content. It is still a powerful soundtrack. The ranking system here means there is little difference between a half star; a 4.25 and 4.75 Star film may both sound just as impactful, with the subtle differences separating them. All ranking systems have quirks, but I think this one is pretty good at capturing great LF content more often than not.

 

JSS

[Bossobass Dave]

-28dB is in reference to the maximum signal possible: all 8 channels with a time coherent 0dBFS signal, adding up to 128dB. So -28dB is still 100dB at the listening position at that single frequency. I use a 2 second FFT in order to have 1/2Hz resolution while not losing transients as much as higher resolutions, and use zero offset. Using offset makes everything look hotter than it actually is, and the higher res the FFT, the larger the time smear.....it's a preference.

 

Our graphs match in relative level for each freq. I use a different maximum so that I never clip, making things look lower in level, but the PvA shows the real deal for each scene... Compared to WotW, TIH and HTTYD, FOTR has less <20Hz content. It is still a powerful soundtrack. The ranking system here means there is little difference between a half star; a 4.25 and 4.75 Star film may both sound just as impactful, with the subtle differences separating them. All ranking systems have quirks, but I think this one is pretty good at capturing great LF content more often than not.

 

JSS

 

 

I understand the quirks in the ranking system. I believe that 5 stars is obviously too rough a gauge. A number scale is prob a better system and would show the fine line differences to a decimal fraction of 1. But, that's entirely your call.

 

I also understand the offset option but I disagree that it makes anything look hotter. The color scale I selected gives around 8dB per color with 18 bars per decade of shading resolution. Changing the offset doesn't affect the levels, it just allows you to use the detail in the selected color scale, which is what the offset option is for.

 

Your graph of the ring scene, for example, shows blue, green yellow and a small portion of red, whereas mine shows blue to a portion of purple. That translates to around 50 bars of shading in your graph vs 90 bars in mine, scene-for-scene.

 

As well, our graphs do not match in relative terms. If I select 8 Hz vs 34 Hz (the hotter blips on your ring scene), You show 8dB difference and I show 3dB difference (approx). If anything, my FR at the LP was higher at 34 Hz than at 8 Hz when I mic'd the scene.

 

I don't understand some of the nuances of the SpecLab program, but with all of the thousands of debates regarding subwoofer systems design and everything related to that over the years, the nuances of SL is not high on the list for me. I'm a bit burnt out on the subject and have slowly taken a more hands off-observer position recently.

 

BTW, along those lines, I'm getting set to wipe the Photobucket account. If you're interested in transferring the pics to another location, ping me. Otherwise, the links will be useless.

[maxmercy]

Changes in FFT settings can account for the relative differences, as a longer FFT has a longer total rise time. Like you said, there is no magic setting, all are compromises. I chose the lowest integration time possible while maintaining enough resolution to avoid blockiness.

 

I would be nice to keep the linked images here, they are invaluable to the thread.

 

JSS

[grantisgrant]

Hey guys, I would be willing to transfer the photobucket images to imgur and provide links. Let me know.  

[maxmercy]

Same Balrog scene, with +20dB offset:

 

Different colors, same data.  

 

JSS

[maxmercy]

 

From my just completed viewing at -25, I would definitely agree with Dave.  Just a few more timestamps, if you'd like to test them out too. From the EE trilogy set

 

Part 1: 

20:30-21:00 First firework goes off

22:50-23:40 Dragon firework

27:40-28:00 Gandalf gets scary

54:25-55:00 (Not sure what this was)

 

Part 2:

9:30-11:00 Rocks and snow fall

17:00-18:30 Kraken thing

38:45-39:30 Stairs collapsing

39:30-41:30 Balrog

 

Thanks! Looking forward to TTT at some points soon.

 

You are gonna have to pick your fav 3 of these....that's a lot of stuff to graph.

 

JSS

[FOH]

The Place Beyond The Pines was pretty good; did not expect it to head the way it did....a good break from the typical Hollywood cookie cutter.

 

JSS 

 

Typical Hollywood cookie cutter fare often produces the meat and potatoes of what threads like this live for. But you're right, it was refreshingly different, and pulled me in and took me on a ride. That's why I threw it out there.  

[grantisgrant]

Dragon Firework, Rocks and Snow Fall, and Kraken thing would be my picks if you tested three. Thanks max(?) I love getting the raw data!

[Bossobass Dave]

Same Balrog scene, with +20dB offset:

 

Different colors, same data.  

 

JSS

 

 

Now you're talkin' 

 

You're now using your entire color graph vs half of it.

 

Taking a little slice and enlarging it, you see an area that seems to have next to nothing going on, or at least what's happening is very hard to discern. The offset adds in the detail that's otherwise missing. This is just a small example. The differences are everywhere in every graph.

 

[MKtheater]

Well I know I am feeling the low stuff during Balrog as The graph shows! I am running the scene at 110 dBs at 5 hz apparently.

[MKtheater]

Double post

[maxmercy]

Now you're talkin' 

 

You're now using your entire color graph vs half of it.

 

Taking a little slice and enlarging it, you see an area that seems to have next to nothing going on, or at least what's happening is very hard to discern. The offset adds in the detail that's otherwise missing. This is just a small example. The differences are everywhere in every graph.

 

There is a problem, though.  With this much offset, you can have greater than 0dBFS, and as much as +20dBFS, and you run out of colors.  It looks like the graph is more 'filled in', but we are talking content encoded at less than 68dB.  20Hz is not even audible until 70dB.  There is reason to my method, I assure you.  Changing things just to see purples and pinks when that stuff is simply not encoded that hot is not my idea of accurate.  On my graphs, yellow means 100dB.  That is pretty damned loud for a single 1/2Hz bin.  Red is >110, and white is >120dB.  Purple should rarely (if ever) show up, and it is that way in order to have enough headroom should it ever occur.

 

JSS 

[maxmercy]

Dragon Firework:

 

Saruman Brings Down the Mountain

 

The Watcher Takes Down the Gate

 

JSS

[FOH]

There is a problem, though.  With this much offset, you can have greater than 0dBFS, and as much as +20dBFS, and you run out of colors.  It looks like the graph is more 'filled in', but we are talking content encoded at less than 68dB.  20Hz is not even audible until 70dB.  There is reason to my method, I assure you.  Changing things just to see purples and pinks when that stuff is simply not encoded that hot is not my idea of accurate.  On my graphs, yellow means 100dB.  That is pretty damned loud for a single 1/2Hz bin.  Red is >110, and white is >120dB.  Purple should rarely (if ever) show up, and it is that way in order to have enough headroom should it ever occur.

 

JSS 

 

Makes sense to me. Granted, many of us, ...myself included, may be more accustomed to the more visually impressive offset. But maxmercy's methodology and accompanying explanation above, makes some good points to me (having never used SL). Yes, getting used to the color scale would just take a bit of time, and both the audibility and headroom elements of the methodology do appear tenable.       

[grantisgrant]

Dragon Firework:

Dragon Firework.jpg

 

Saruman Brings Down the Mountain

Saruman Brings down the Mountain.jpg

 

The Watcher Takes Down the Gate

The Watcher akes down the Gate.jpg

 

JSS

 

Thanks I figured they wouldn't be that low, just wanted to see exactly how low!

[Bossobass Dave]

There is a problem, though.  With this much offset, you can have greater than 0dBFS, and as much as +20dBFS, and you run out of colors.  It looks like the graph is more 'filled in', but we are talking content encoded at less than 68dB.  20Hz is not even audible until 70dB.  There is reason to my method, I assure you.  Changing things just to see purples and pinks when that stuff is simply not encoded that hot is not my idea of accurate.  On my graphs, yellow means 100dB.  That is pretty damned loud for a single 1/2Hz bin.  Red is >110, and white is >120dB.  Purple should rarely (if ever) show up, and it is that way in order to have enough headroom should it ever occur.

 

JSS 

 

 

Yeah, that's (+20dBFS) not gonna happen.   Even if that would ever happen, again, you have the offset option (no one said 20dB offset was a rule).

 

Regarding accuracy, that's what the peak hold, average and "peak at" functions are for. Even then, how accurate is your calibration?

 

The waterfall graph is for viewing the content, not reading numbers. Losing detail for the sake of closer accuracy, especially this far into the game, makes little sense to me. I'm not interested in seeing what content is audible by conventional wisdom or some other irrelevant metric. I want to 'see' what's there, in as much detail as possible.

 

If you prefer the 'closer to the recorded numbers' approach in a waterfall, you should adjust your scale range and colors accordingly. My point is that there's no point in having a 65dB range with 117 bars of resolution if you're only using 1/2 that or less.

 

Of course, it's your call. 

[MKtheater]

I see the OHF peak hold graph hit 110 dBs at 1 hz.  I see about 19hz hit 115 dBs, where does the 124 dBs come from?  If we truly are hitting 124 dBs from bass management then if you want the waterfall to correspond to that it should be purple, no?

[lfe man]

As requested:  Lord of the Rings: Fellowship of the Ring - Extended DVD Edition DTS-ES 6.1:

 

Level - 4 Stars (112.2dB composite)

Extension - 3 Stars (16Hz)

Dynamics - 5 Stars (29.3dB)

Execution - Will poll

 

Overall - TBD

 

Recommendation - TBD

 

This mix is 3dB hotter, and 2dB more dynamic than the BD Theatrical mix.  It appears to be a better mix overall.  

 

The good news is that the BD Extended Edition Mix is nearly identical to this one, with a very similar PvA.

 

So, the BD and DVD Extended Editions appear to be at least more Dynamic (that could be because more dialogue), and have more Level than the Theatrical BD. 

 

But the extension is identical to the BD Theatrical Release and BD Extended Edition, we seem to remember more extension in our audio memories.

 

I feel no need to do the other DVD Extended Editions since the BD Extended Edition was so close to its DVD Extended Edition counterpart for this film.  

 

EDIT - There is very little clipping in this mix.  What there is, only hits 0dBFS for one or two samples, then comes back down.  Very well executed.

 

 

JSS

Thanks, though dolby track might be hotter tham dts track what i remember.

[maxmercy]

Yeah, that's (+20dBFS) not gonna happen.   Even if that would ever happen, again, you have the offset option (no one said 20dB offset was a rule).

 

Regarding accuracy, that's what the peak hold, average and "peak at" functions are for. Even then, how accurate is your calibration?

 

The waterfall graph is for viewing the content, not reading numbers. Losing detail for the sake of closer accuracy, especially this far into the game, makes little sense to me. I'm not interested in seeing what content is audible by conventional wisdom or some other irrelevant metric. I want to 'see' what's there, in as much detail as possible.

 

If you prefer the 'closer to the recorded numbers' approach in a waterfall, you should adjust your scale range and colors accordingly. My point is that there's no point in having a 65dB range with 117 bars of resolution if you're only using 1/2 that or less.

 

Of course, it's your call. 

There is no calibration.  I am reading straight off the disc, and using accepted standards (LCRS 105dB sinewave RMS peaks, LFE 115dB sinewave RMS peaks).  Put all 7.1 channels together and that is 128dB (not quite, but within 0.1dB) absolute peaks, with a sinewave in each channel, 125dB RMS.  

 

I 'bass manage' the data itself, so that if it reaches 0dBFS, I know it is a 128dB event...and some films come damned close, but not at one frequency.

 

I could re-arrange the color scheme, or just shift it or squeeze it some, so that more colors are displayed.  But IMO, purple should be a VERY rare event, and when it is seen, it should completely grasp your attention.

 

I can get rid of the superfluous 5dB on the top, that should help things out.

 

How about this:

 

THX Amazing Life Trailer:

 

Now the white/purple means something; that little part of the freq spectrum is encoded at ~118dB!!!  There are films that come closer than you think....

 

 

JSS

[maxmercy]

I see the OHF peak hold graph hit 110 dBs at 1 hz.  I see about 19hz hit 115 dBs, where does the 124 dBs come from?  If we truly are hitting 124 dBs from bass management then if you want the waterfall to correspond to that it should be purple, no?

 

The overall peak is the sum of all frequencies in the loudest instant of the film.  It is rarely all in one frequency.  You have to 'add' all the bins to get the total.  The peak is the peak is the peak dB.  The more resolution in an FFT, the 'lower' each 'pixel' will be encoded.  That's what I meant by tradeoff; the more resolution you have, the longer an effect needs to 'build' to show its true level.  Bosso adjusts for it with the offset command.  I try to make everything apples/apples knowing that by increasing resolution, I decrease each bin's level, as they must all sum to the same 'total'.....it's actually more complicated, but that's the easy way to explain it...

 

JSS

[maxmercy]

Thanks, though dolby track might be hotter tham dts track what i remember.

 

Nope, it has dialnorm at -27, so it will be 4dB softer, IIRC.

 

JSS

[Bossobass Dave]

There is no calibration.  I am reading straight off the disc, and using accepted standards (LCRS 105dB sinewave RMS peaks, LFE 115dB sinewave RMS peaks).  Put all 7.1 channels together and that is 128dB (not quite, but within 0.1dB) absolute peaks, with a sinewave in each channel, 125dB RMS.  

 

I 'bass manage' the data itself, so that if it reaches 0dBFS, I know it is a 128dB event...and some films come damned close, but not at one frequency.

 

I could re-arrange the color scheme, or just shift it or squeeze it some, so that more colors are displayed.  But IMO, purple should be a VERY rare event, and when it is seen, it should completely grasp your attention.

 

I can get rid of the superfluous 5dB on the top, that should help things out.

 

How about this:

 

THX Amazing Life Trailer:

THX AL.jpg

 

Now the white/purple means something; that little part of the freq spectrum is encoded at ~118dB!!!  There are films that come closer than you think....

 

 

JSS

 

 

Yes, that's what my opinion is. If you squeeze the colors into the actual window when set at '0' offset, it's the same difference as using the offset and leaving the scale as-is.

 

BTW, as you know, I mic the subs for the data to generate the SL graphs, that's why I have the +5dB on my scale. Running the subs hot and/or a peak in response that coincides with a particular effect when casually graphing or just seeing what the limits of the system are requires it. I recall MKT talking about "pegging the meter" as he runs up to +10dB hot often.

 

In any case, FR non-linearity and less-than-calibrated system (because I move stuff in and out very often and sometimes don't have it fine-tuned calibrated, or I set the mic down in a random spot and graph while I watch the flick are the reasons I'm not overly concerned with exact waterfall calibration.

 

Also, while you're on the subject, exactly how does SL read straight off the disc? It can't decode the formats, so how do you inject the data into SL directly?

 

I don't see how SL reads directly off the disc.

[Bossobass Dave]

 

 

I 'bass manage' the data itself, so that if it reaches 0dBFS, I know it is a 128dB event...and some films come damned close, but not at one frequency.

 

Now the white/purple means something; that little part of the freq spectrum is encoded at ~118dB!!!  There are films that come closer than you think....

 

 

JSS

 

 

 

I understand the theory, but show me a movie that has coherent content at 0dBFS in every channel at once. If it existed, it would be the talk of the town. There's 10 times the difference between 118dB and 128dB.

 

BTW, I get 125.2dB for 7 channels @ 105 + 115 LFE channel, of coherent sound.

[maxmercy]

RMS. Add 3dB for peak readings, and you hit the 128dB mark. I do not have an example film, but given the loudness war we have seen thus far, would you not be surprised if a128dB peak shows up? A few films get above 125dB peak, they just don't do it at one frequency. Like I said, I am prepared for the worst case scenario. The Audio Test BD I am working on will contain many such worst case scenarios...

 

JSS