Print

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

[Matt]

Setting the levels in DSP Studio > Room Correction is tricky for a few reasons:

1) You must set distances first since this can effect levels for matched speaker sets

2) You must pick 'Solo' and 'Tone' to get a tone from one speaker; picking 'Solo' on one speaker doesn't disable 'Solo' on other speakers

3) Increasing a level could actually just decrease all the other speakers making previous measurements invalid (reducing other speakers is technically correct, as it's done to avoid the possibility of clipping)

4) The 'Tone' isn't a very good way to dial in a subwoofer level because, at least with my subwoofer, the decibel level is much higher with a pink noise signal than with a pink noise signal with a 100 or 50 Hz lowpass applied

Someday I would like a wizard (possibly with a microphone) to walk through this, but that won't happen for v16.  So I'm trying to think of simple solutions to the problems above.

One solution for #2 and #3 that wouldn't require much new code would be to add a checkbox under 'Tools' for each speaker that would be called 'Level'.  Checking it would play a tone from just that one speaker.  The tone would disable the clip prevention, so turning up a level slider would always make that level higher.

I'm not sure what the best type of level tone would be for the subwoofer.  I know you can do full sweeps with a calibrated microphone, but that's a little beyond what I'm looking for.  Maybe just adding a low pass filter to the tone at 60 Hz for the subwoofer tone?

For #1, simply adding a note to the text on the dialog might be a good start.  Or changing the order so 'Distance' comes before 'Level'.

Thanks for any tips.

[mojave]

1.  The distance should be set first. Moving it to the top and having some text about it is a good idea.

2.  I have found it easier to use a calibration DVD played through media center to set the levels. The DVD will play a tone from each speaker without one having to click on anything. You can leave the SPL meter in the same place and record the level for each speaker. The tone method in MC is very awkward and tedious to use. It would be better if MC played a tone through each speaker the same way. You could have a calibration button that would play a tone through as many speakers are set in Output Format. One could write down the SPL level for each speaker. It would be nice if we could then enter this level in MC for each speaker and it would automatically adjust everything.

3.  Whenever I've done calibration/setup using the soundcard drivers or a calibration disc there has been a visual image of all speakers and the sub. Clicking on each one would produce a test tone from that speaker. This was helpful for quickly checking connections and levels.

4.  Maybe you could add a Setup tab to Room Correction that incorporates #2 and #3. It could have a start button and a finished button. The start button activates the tones that rotate from speaker to speaker. After one records all the SPL levels, the finished button would be clicked. MC would then calculate all the levels and set the speakers to the correct level. Clicking on each speaker's tab would allow one to manually adjust the levels.

5.  Regarding the subwoofer test tone, I think it should match the crossover frequency the user has set in bass management. Whenever one does calibration with REW, the signals are always going through the crossover settings for an accurate reading. A subwoofer isn't calibrated apart from bass management.

[mojave]

Since Room Equalization Wizard is such a powerful measurement program I think it would be worth it to spend time integrating it somehow into MC (as a DSP). I read somewhere that John M., the developer, hopes to add multi-channel measuring to it so one could use all their soundcard outputs instead of just one channel. If this happens, one could easily use it to check their distance settings, levels, and run sweeps. REW could then generate a calibration file that MC would automatically load which would adjust Room Correction and the Parametric Equalizer. It would be calibration heaven! I think that adding a way to use REW with MC would add many customers from Home Theater Shack, the home of REW.

I've never tried it before, but many also like the HOLMImpulse software. It is similar to REW and could possibly work with MC. The freeware license says, "If you'd like to use the product in a commercial way contact HOLM Acoustics (info@holmacoustics.com)."

[Matt]

I need to make sure I have my thinking right on this.  For the sake of my example, assume a 5.1 setup.

I believe a pink noise tone on any channel should produce the same sound level (and the subwoofer should be +10 dB).  Pink noise is good because it has equal energy at all frequencies.  Adding a low or high pass filter to pink noise shouldn't change the sound level as measured by a meter.

I can imagine two main ways to set the speaker levels (and in particular the subwoofer level):

1) Play a 5.1 pink noise file with a subwoofer at -10 dB.  Disable bass redirection. Adjust each speaker so it's the same volume.

or

2) Play a 5.1 pink noise file with a silent subwoofer.  Enable bass redirection.  Adjust each speaker so it's the same volume.

Is one of these a better practice than the other?  #1 seems pure because it's simply saying "A full volume signal will always produce x decibels of output."  #2 pushes the signal through the filters used during playback, so it's more of a real-world calibration (but I'm not certain the level from this would match the correct level during true 5.1 playback where bass redirection is often adding little bass).

Thanks.

[mojave]

I don't think one is better than the other. If the bass management is summing correctly won't both methods yield the same result?

[mojave]

I just purchased the MiniDSP which has Auto-EQ tuning with REW. I've been wanting one for testing and messing around. I'll probably use it to play with various EQ settings with REW and then input the final settings into MC's Parametric EQ and take the MiniDSP back out of the signal chain. I'll only be using it for the subwoofer.

At the end of the MiniDSP setup page it says, "The whole miniDSP team would like to thank John Mulcahy, Author of REW, for working along us in providing this neat feature." Maybe MC will be able to say something similar someday. 

[mojave]

Version 16.0.82

What is the purpose of having both a "Level" and a "Tone" button? I think renaming "Level" to "Tone" and getting rid of the current "Tone" makes more sense. Why would I want pink noise in one speaker and music playing in the rest (which the current "Tone" does)? Also, the term "Level" doesn't convey what the button does. The button doesn't level anything, it just plays a tone.

A more important issue is the requirement that music must be playing in order to calibrate using the pink noise. If I want the test tone, I should just be able to click the button and the test tone will play. The test tones shouldn't be dependent on the source number of channels either. A user might like to use "Source number of channels" in Output Format and no JRSS, yet still have a 7.1 system. How is he supposed to calibrate his system? He has to find a source with 7.1 channels and play it while pressing the test tone button in Room Correction.

Another nice option would be "Play a tone for X seconds from each speaker." Even though the SPL meter might say the rear surrounds are level matched to the mains, when actually listening they might sound quieter due to my ears facing forward. I like to be able to hear the tones sequentially as a final check of things.

[Matt]

What is the purpose of having both a "Level" and a "Tone" button? I think renaming "Level" to "Tone" and getting rid of the current "Tone" makes more sense. Why would I want pink noise in one speaker and music playing in the rest (which the current "Tone" does)? Also, the term "Level" doesn't convey what the button does. The button doesn't level anything, it just plays a tone.

I didn't want to take functionality away.  I could imagine a case where you might want a tone in two or three speakers to measure the total level.

A more important issue is the requirement that music must be playing in order to calibrate using the pink noise. If I want the test tone, I should just be able to click the button and the test tone will play. The test tones shouldn't be dependent on the source number of channels either. A user might like to use "Source number of channels" in Output Format and no JRSS, yet still have a 7.1 system. How is he supposed to calibrate his system? He has to find a source with 7.1 channels and play it while pressing the test tone button in Room Correction.

This is a limitation of doing the work in a DSP.  I agree it's not perfect, but it's too complicated to change for v16.

Another nice option would be "Play a tone for X seconds from each speaker." Even though the SPL meter might say the rear surrounds are level matched to the mains, when actually listening they might sound quieter due to my ears facing forward. I like to be able to hear the tones sequentially as a final check of things.

Good idea.

[mojave]

This is a limitation of doing the work in a DSP.  I agree it's not perfect, but it's too complicated to change for v16.

Here is a different way of thinking about it, but I don't know if it would work. You could include a 7.1 pink noise wav file with MC. Starting calibration could load the wav into Playing Now and start playing it (with Repeat Current Item on). It would then go through Output Format so that things like phantom center could also be tested/checked. The wav file would be muted until "Tone" was clicked which would essentially unmute it for just that speaker. If you wanted to hear it in several speakers at once, you could check "Tone" for each speaker.

[Matt]

Here is a different way of thinking about it, but I don't know if it would work. You could include a 7.1 pink noise wav file with MC. Starting calibration could load the wav into Playing Now and start playing it (with Repeat Current Item on). It would then go through Output Format so that things like phantom center could also be tested/checked. The wav file would be muted until "Tone" was clicked which would essentially unmute it for just that speaker. If you wanted to hear it in several speakers at once, you could check "Tone" for each speaker.

Creating a tool that would make a set of test tone files would be a nice addition to the program, I think.  Pink noise isn't compressible, so it's better to build the files than to include or distribute them.

[mojave]

I like today's changes with 16.0.84. I'm going to try to run through a calibration at home this weekend with my SPL meter.

I'll probably start using JRSS for music more with the change to Room Correction's bass management. If I use "Move center to front L/R" and JRSS, do my mains have the original signal or is the JRSS created center folded back into the mains?

[Matt]

If I use "Move center to front L/R" and JRSS, do my mains have the original signal or is the JRSS created center folded back into the mains?

Moving the center has no impact on a JRSS created subwoofer, which seems correct to me.

[Matt]

I like today's changes with 16.0.84. I'm going to try to run through a calibration at home this weekend with my SPL meter.

I'm going to write this up once I'm happy with how it works, but it should be pretty simple now:

1) Set speaker distances in DSP Studio > Room Correction
2) Set whether your subwoofer is +10 dB in DSP Studio > Output Format (it should be in most cases)
3) Play some content that uses all your speakers
4) In DSP Studio > Room Correction, go through each speaker and click the 'Level' tool.  Move the volume slider on the same dialog until the SPL meter matches the other channels

[mojave]

Moving the center has no impact on a JRSS created subwoofer, which seems correct to me.

Unlike many of my posts, I didn't say anything about the subwoofer.  I want to know if the mains are "unmolested" by JRSS when using a phantom center. Does JRSS matrix a center and that selection folds it back into the mains, or does JRSS not create a center at all?

[Matt]

Unlike many of my posts, I didn't say anything about the subwoofer. 

It's like Simon Says

I want to know if the mains are "unmolested" by JRSS when using a phantom center. Does JRSS matrix a center and that selection folds it back into the mains, or does JRSS not create a center at all?

It folds it back in, but I'll change it next build.  This is just a corner case.

[mojave]

I used the Level button and viewed the audio spectrum through several spectrum analyzers and compared results using the left front channel. With both Gliss EQ and Blue Cat's Audio, the frequency response follows a 3 dB/octave drop off. With Analyzer, the frequency is flat. I then checked the pink noise tests you had sent me a while a go and discovered the same thing. The pink noise test files also follows a 3 dB/octave drop off, but shows as correct in analyzer. My conclusion is that the pink noise and Analyzer are both off by 3 DB/octave slopes, but in opposite directions so that a flat frequency response is the result.

Gliss EQ (and SPAN) allow you to adjust the slope of the frequency response. The default for Gliss EQ is 4.5 db/Octave and I think the default for SPAN is 3 db/octave. The reason is that this helps when mastering to follow the Flechter-Munson curves. A sound mixer will change the slope of the frequency response and then mix so it looks flat on the screen. In reality, the audio is increasing by 3 dB or more per octave as the frequency decreases. Blue Cat's Frequency analyzer is flat.

I'm wondering if you orginally created flat pink noise files bases on SPAN (and its default setting). This would have resulted in incorrect slopes. If you then calibrated Analyzer to the pink noise it would still be incorrect. This is just a guess, though. Maybe I'm looking at this wrong.

When using the Levels, it would be nice to be able to cycle through speakers using the arrow keys/buttons. It is hard to see the little box when sitting 10' ft from a 24" LCD and I didn't want to turn the projector on.

[Matt]

I think pink noise is supposed to have a 3 dB/octave rolloff.

For example, download the clip and spectrum here:
http://www.dogstar.dantimax.dk/testwavs/

Our analyzer is configured so that pink noise looks flat.  This is like picking a 3 dB/octave slope in SPAN.

[mojave]

I tested Gliss EQ this afternoon with the DIY Test DVD from AVS. It has test tones that start at 1 Hz and go up to 120 Hz (almost 7 octaves). It increases by 1 Hz from 1 to 120 Hz and then by 5 Hz from 20-120 Hz. I played them up to 100 Hz with Gliss EQ  set to MAX to see if it was showing a flat frequency response (with the slope at 0). It was and is extremely accurate in showing the test tones. If Analyzer had a max hold capability, it would have been down 18 dB by 64 Hz.

There is probably a lot of pink noise that is available with a 3 db/octave slope since this is the standard for mastering. With playback, though, we want to see the actual levels. I also used the pink noise on the Levels section of the DIY Test DVD and that pink noise is perfectly flat. I doesn't really made sense to use sloping pink noise when doing a levels test. You basically end up measuring the SPL at the lower frequencies since they will be the loudest.

Today I ordered the Blu-ray Calibration Disc from AIX Records. I'm going to check the channel output and levels with it, too, and compare to MC.

[Matt]

I think the definition of pink noise is basically white noise that falls off by 3 dB/octave.

Look at the graphs here (the one of pink noise mostly matches the Blue Cat graph above when it's configured to have no slope):
http://en.wikipedia.org/wiki/Colors_of_noise

[Matt]

It would be interesting to try white noise (you could make samples with Media Editor), and see what that looks like in Blue Cat.

[mojave]

I didn't know pink noise had a 3 db/octave slope. Here is white noise. It plays flat in Blue Cat and Gliss EQ. Maybe the Levels button should play white noise instead of pink noise and Analyzer should play flat with white noise.

[mojave]

This thread has a lot of information about Understanding and Using Test Tones. I'll read it when I get home.

[Matt]

I didn't know pink noise had a 3 db/octave slope. Here is white noise. It plays flat in Blue Cat and Gliss EQ.

Thanks for the tests.  I think they show that:

1) Test tones in Media Center are proper pink noise
2) DSP Studio > Analyzer shows pink noise as a flat line

So both of these are functioning as designed, which is good.  You had me nervous for a while

Maybe the Levels button should play white noise instead of pink noise and Analyzer should play flat with white noise.

I think pink noise is the standard for level calibration.  It makes sense to use a logarithmic scale (octave-based) because that's how we hear.

There may be times when you want a full amplitude sine wave at a certain frequency or white noise. 

One addition I think would be neat in Media Center would be a "Create Test Clips..." function that would make a pile of WAV files for all sorts of testing and tuning purposes -- things like tones, sweeps, 5.1 tests, 7.1 tests, etc.  Test files would be big to download (noise can't be compressed) but easy to generate, so a tool to generate the files makes the most sense to me.

[Matt]

Here's a snippet from the link you posted:

On a logarithmic frequency plot, pink noise is represented by a flat horizontal line.

This is agreeing with how Analyzer works.  Analyzer uses a logarithmic frequency plot along the x-axis, and shows pink noise as a flat horizontal line.

I think the configurable slope in SPAN is a little confusing.  I would expect the x-axis to change to linear (from logarithmic) if you wanted to see white noise as a flat line.

[mojave]

You had me nervous for a while 

Sorry. What started me down this path is I'm trying to figure out why the movie frequency plots posted on AVS don't seem to match what I see in MC with spectrum analyzers. I then noticed the analyzers were different.

[mojave]

I changed the slope back to 3.0. Look at the massive spike at 3 Hz on How To Train Your Dragon when the big dragon is killed.

[Matt]

I changed the slope back to 3.0. Look at the massive spike at 3 Hz on How To Train Your Dragon when the big dragon is killed.

I'm not sure the frequency numbers down that low are super reliable.

Let's say you do a 4096 sample FFT (SPAN may default to less, but I think this is what Analyzer uses), you get 2048 frequency points.

Assuming you're playing at 48 KHz, each FFT point then spans 23 Hz.

So point 0 is the overall level (ignored), then point 1 is from 1 Hz to 23 Hz, point 2 is 24 Hz to 46 Hz, and so on.

The entire subwoofer span is only a few frequency points, so it doesn't lend itself to very fine analysis.  I suppose this is why test tones that only play a certain frequency are valuable for a subwoofer.

[Matt]

Since, for the reasons I explained above, an FFT probably won't spike at 4 Hz and fall at 10 Hz unless it's using a huge window size, it's possible that SPAN incorrectly incorporates point 0 coming out of the FFT in the low frequencies.

The first data point out of an FFT is 0 Hz which is just the overall level.

[mojave]

I'm not sure the frequency numbers down that low are super reliable.

Let's say you do a 4096 sample FFT (SPAN may default to less, but I think this is what Analyzer uses), you get 2048 frequency points.

That is why I tested the test tones from 1 Hz to 100 Hz. I didn't show a chart yet of the individual tones, but just the max output from 1 Hz to 100 Hz. In that chart you can clearly see the peaks at 25, 30, 35 Hz etc where the test tones played. I'll post an individual tone tomorrow.

I was using a 32,768 sample FFT in Gliss EQ.  How many Hz does that span?

[mojave]

Here are 4 Hz and 20 Hz test tones. It looks like the spectrum analyzer is accurate with 32,768 samples.