INTERACT FORUM

Please login or register.

Login with username, password and session length
Advanced search  
Pages: 1 2 [3]   Go Down

Author Topic: Bass Management Testing  (Read 46245 times)

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Bass Management Testing
« Reply #100 on: August 06, 2010, 05:24:59 pm »

Let me expand. My question was that if you started with the LFE at 115db and the other channels at 105dB, would it sum to the same totals (based on number of channels) that soho54 got using the calculations that MC uses? If I am playing back 7.1 in MC and everything is at full signal strength (ie 115 & 105 dB), then is the combined LFE + redirected bass output going to equal around 117.28 dB? This is assuming the bass is slightly different in each channel

I'm glad the post was helpful. Now that the extensive bass management options are available in the public build, I'm thinking about starting a post at AVSForum's HTPC subforum.
Logged

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42323
  • Shoes gone again!
Re: Bass Management Testing
« Reply #101 on: August 06, 2010, 05:33:22 pm »

Let me expand. My question was that if you started with the LFE at 115db and the other channels at 105dB, would it sum to the same totals (based on number of channels) that soho54 got using the calculations that MC uses? If I am playing back 7.1 in MC and everything is at full signal strength (ie 115 & 105 dB), then is the combined LFE + redirected bass output going to equal around 117.28 dB? This is assuming the bass is slightly different in each channel

With build 15.0.92 or newer, the levels will match exactly.

Of course, there may be attenuation to avoid possible clipping, so instead of the LFE getting a 2.28 dB boost, the other channels get a 2.28 dB cut.

But remember that the notion of using "115 dB" and "105 dB" are sort of arbitrary.  The math would work out the same if you used 60 dB and 50 dB.
Logged
Matt Ashland, JRiver Media Center

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Bass Management Testing
« Reply #102 on: August 06, 2010, 06:28:11 pm »

Wow, that was quick!
Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14401
  • I won! I won!
Re: Bass Management Testing
« Reply #103 on: August 13, 2010, 07:01:13 am »

How does the Crossover setting for the Subwoofer work, does it:
1) Cut everthing under XXhz over to the 0.1 channel (eg remove from L/R)
2) Add everthing under XXhz to the 0.1 channel  (eg keep on L/R as well)

I like the option you get in FFDSHOW to chose the behaviour (and I personally like keeping the base on the main channels as well - I think you get a better integrated bass timber over the front sound stage)

Thanks
Nathan
Logged
JRiver CEO Elect

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42323
  • Shoes gone again!
Re: Bass Management Testing
« Reply #104 on: August 13, 2010, 08:47:58 am »

How does the Crossover setting for the Subwoofer work, does it:
1) Cut everthing under XXhz over to the 0.1 channel (eg remove from L/R)
2) Add everthing under XXhz to the 0.1 channel  (eg keep on L/R as well)

The subwoofer crossover you select in DSP Studio > Output Format is used when simulating a subwoofer channel for sources that have no LFE channel of their own.  For example, it comes into play when mixing 2.0 to 5.1 with JRSS.

It makes no changes to other channels.


Quote
I like the option you get in FFDSHOW to chose the behaviour (and I personally like keeping the base on the main channels as well - I think you get a better integrated bass timber over the front sound stage)

JRiver offers quite a lot more than FFDShow here.  The key is that to do bass redirection (i.e. moving bass from other channels to the subwoofer), you use the 'Room Correction' DSP.  We allow crossover frequency per speaker, and also configurable slope for the high and low pass filters.
Logged
Matt Ashland, JRiver Media Center

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Bass Management Testing
« Reply #105 on: August 13, 2010, 09:36:09 am »

There are currently two DSP's that have bass management settings:  Output Format and Room Correction.

Output Format
Like Matt mentioned, this setting only comes into play when you don't already have any information in the subwoofer channel. This would be the case if you are upmixing 2.0 to 5.1. If you already have content, like with a 5.1 or 7.1 source, then this setting is ignored. It uses a 48 db/octave low pass filter for the crossover frequency. It only copies the bass, but it doesn't remove it from the other channels.

Room Correction
This is handles bass management in two situations:  1)  when you want to remove bass from other channels for content upmixed from 2.0 to 5.1 or 7.1 and 2)  when you want to perform bass management on 5.1 or 7.1 sources.

This gives the option of crossover frequency, high and low pass filter slopes (they aren't labeled, but the top one is high pass filter for each channel and the bottom one is the low pass filter for the subwoofer channel), and whether the bass is removed (really only useful for upmixed sources), moved, or copied.

The Conflict
First, people see bass management settings in Output Format and don't realize that the settings that actually affect 5.1 or 7.1 content are under Room Correction.

Second, if Room Correction is set for Move Bass, like I assume most will use for 5.1 or 7.1 content, then these settings won't work right for upmixed 2.0 content. I suggested a method of integrating both settings earlier in this thread.
Logged

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Bass Management Testing
« Reply #106 on: August 13, 2010, 11:25:31 am »

Here is what the Bass Management in Room Correction looks like:

Logged

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14401
  • I won! I won!
Re: Bass Management Testing
« Reply #107 on: August 13, 2010, 09:34:03 pm »

So much to play with  ;D

Anyway I "think" I am correct in saying from reading the above that
1) 2.0 Music --> 5.1 with base being "copied" to LFE (say a 85Hz Crossover with a 48db/ocatve low pass filter) is all done in Output Format by setting JRSS with my 2.0 Music I can get Base
2) 5.1 is then left as is
3) I can ignore Room Correction

If so I'm happy with that (see pic).

I'm still using FFDSHOW for both 2.0 --> 5.1 mixing in and LFE'ing in Video Sources till the JR Audio Renderer supports (and hence I can replace Reclock):
- Slave Audio to Video Clock, and (to a lesser extent)
- Change refresh rate to match Video frame rate

Thanks
Nathan
Logged
JRiver CEO Elect

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42323
  • Shoes gone again!
Re: Bass Management Testing
« Reply #108 on: August 13, 2010, 11:12:10 pm »

- Slave Audio to Video Clock, and (to a lesser extent)

Would you mind starting a thread on this and explaining (again)?

There's really no such thing as "slaving".  There is one filter in charge of the clock, but it's normally run as a system clock (we experimented with running it based on the sound card clock and it just doesn't work).

The audio render has to manage drift between the audio and video.  I suppose it could use a rate adjuster for the audio signal (we already have this in our DSP Studio) to change audio playback rates slightly on the fly to account for drift instead of issuing corrections that the video has to sync to.

But I would like to understand if this is a real world problem or a theoretical problem.  I've never seen a stutter during video playback.  Have you tried the JRiver audio renderer?  Are you actually seeing stutters?

Thanks.

Logged
Matt Ashland, JRiver Media Center

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14401
  • I won! I won!
Re: Bass Management Testing
« Reply #109 on: August 14, 2010, 12:37:32 am »

Matt - I've attached the ReadMe doco from Reclock, and while it is old the first few pages describe the various issues faced with Video playback from a PC better than I could.   I use the term "slave" poorly but it is really about dynamically adjusting the audio sampling to keep in track with the video playback (rather than the other way around and as happens in CE Players).  Like all "video/audiophile" things it is real for "some", but I'm sure most are oblivious to it all (and certainly for NTSC users you probably think it is normal as you have been consuming 3:2 pulldown material for ever, just like for us PAL users we are used to the 4% pitch increase on film based material).  It is a complex interaction between all the parts of the chain and sometimes it can not be solved.  My Intel G33 @ 24hz would repeat a frame every 17sec (about 1 per 400 frames) regardless of what I did.  Once I noticed it, I was gone....I'd sit there counting down to the next one (very off putting when trying to be absorbed in the content).  In reality we have 48,000 to 192,000 audio samples per second to play with yet as low as 24 for the matching video samples, so any dropped or repeated video Frame is going to be much more noticeable than resampled audio. 

The reason I bring it up again, is that JR is spending alot of time and effort to ramp up the pure Audio side of the house for audiophile consumers and I thought I'd raise the focus on the video side as well, where I think we have:
- Sinc Audio playback to Video Clock,
- Auto change refresh rate to match Video frame rate
- Support high quality Video Renderers (eg madVR has never worked through "support" was added)
- I'm sure there are others!

I'll ask others far more knowledgeable than me to chip in.

Thanks
Nathan

PS - I'll let you split off this and the previous post to a new thread as I can not.
Logged
JRiver CEO Elect

jmone

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 14401
  • I won! I won!
Re: Bass Management Testing
« Reply #110 on: August 15, 2010, 02:31:48 am »

OK - I did some testing with an MKV I made of Shooter (good opening scenes with lots of panning that shows up any inserted or dropped video frames as a stutter).
1) JR Renderer @ 50hz (my std refresh rate being in a PAL country) = Crap, lots of telecine judder
2) JR Renderer @ 24hz = Pretty good but about every 45 Sec I'd see a stutter during pans (not noticeable on still shots)
3) Reclock @ 24hz = Not stutter

Lets look at what is happening (see the reclock pic attached).
- Movie is encoded at 23.976fps (original was film so it is 24hz but tweaked for you NTSC countries)
- The actual refresh rate of my Video Card is 24.002hz
- Difference is about 0.1% or it is One Frame out of sync with the Audio each 42 Seconds (confirms what I see)

While the iPod'ers may not notice, imagine the uproar if there was an audio stutter every 45seconds from your audiophile crowd!  It is no different on the Video side, once you notice it the stutter really gets under you skin...

You can see what Reclock is doing.  It has increased the audio sampling from 48000hz to 48048 (or about 0.1%) so it is now in time with my Video Card clock of 24.002hz (which is 0.1% faster than the video).  Reclock with keep changing this sampling rate so that that the Video Renderer neither needs to add or drop a frame.

Now one frame every 45 sec is not a big thing for most but if MC is to be a audiophile/videophile product such correction will be a plus for some.

Matt - I know you have golden ears but are not sure if video stutter is real so I'd suggest as a test:
1) Use a decent size TV that support 60hz, 50hz and 24hz
2) Play a (blu ray rip) with a good panning section at 60hz, 50hz and 24hz using the JR Renderer and see how the video flows
3) Do the same but this time with reclock

Hopfully your golden eyes will now see the stutter in poorly matched video timing!

I see that there are two potential options in MC:
1) Duplicate all the audio resampling & refresh rate changing stuff that Reclock does into the JR Audio Renderer or
2) Change the JR Audio Renderer so it is a transform filter and can still connect to the Reclock renderer.  This way we can still use all of the DSP stuff you have made in MC + the stuff reclock already does so well.

Thanks
Nathan
Logged
JRiver CEO Elect
Pages: 1 2 [3]   Go Up