Print

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

[Matt]

Would someone with a well calibrated system and a consumer level decibel meter (like the common Radio Shack model) test that the output from Tools > Advanced Tools > Audio Calibration... > Volume calibration produces level readings that match between channels, including the subwoofer?

I want to document this model and make it the recommended approach for level calibration (unless you use convolution and a microphone), but I want to be certain we have everything correct (+10dB for subwoofer, reasonable frequency range for the narrow-band noise, etc.)

Thanks for any help.

[bobkatz]

Would someone with a well calibrated system and a consumer level decibel meter (like the common Radio Shack model) test that the output from Tools > Advanced Tools > Audio Calibration... > Volume calibration produces level readings that match between channels, including the subwoofer?

I want to document this model and make it the recommended approach for level calibration (unless you use convolution and a microphone), but I want to be certain we have everything correct (+10dB for subwoofer, reasonable frequency range for the narrow-band noise, etc.)

Thanks for any help.

I'd really like to help, Matt. Unfortunately you caught me with my pants down  :-(. I had to give up on surround temporarily while I got stereo going with Acourate Convolver...  Now that I'm past that I'm ready to "rebuild" the surround system. But as you can see in my other post of today I want to get a low-latency solution going with JRiver for video playback. And I can't seem to figure out if JRiver can handle stereo subs with its dsp tools.

If we can conquer that I'll be fully calibrating the system with sweeps and REW (cross checked with FuzzMeasure and Audio Toolbox) and my Josephson C550. Finally I'll be able to check out JRiver's own narrow-band PN as you ask with my Rat-shack (analog style) meter which has been calibrated with a calibrator.  (Whew, that was a long sentence)!

So sorry you caught me at this state.... I really want to help.

[bobkatz]

By the way, I meant to report that after accurate convolution/correction with AcourateConvolver, which extended my system to flat below 20 Hz, the measured SPL level difference between full range pink noise at -20 dBFS RMS* and narrow band pink noise 500-2k also set to -20 dBFS is now less than 0.5 dB. So it's just a confirmation of the adage that the flatter and more extended your system is, the closer the 500-2k will get to the full range. In the past, I've seen differences of 1 dB or larger.

Point being that the less confident you are of your system, better to use narrow band pink noise.  HOWEVER, and this is a big however, there was a time when my left front loudspeaker was a little low at 400 Hz and a little high at 1 kHz compared to the right speaker, so narrow band pink noise actually produced a greater measurement error than wideband! After convolution correction with Acourate, the interchannel correlation went up so much that these errors have really disappeared.

As you can see, there are so many potential sources of error that the deeper you dig, the more you have to take into account.  I think it is reasonable to say that someone with a Rat-Shack meter and uncorrected loudspeakers that are in good shape and a good brand can get within, say, 1 dB of correct if he is careful using the narrow band PN you will be supplying.

* (ref. 0 dBFS peak = 0 dBFS RMS with sine wave)

[skeeterfood]

Would someone with a well calibrated system and a consumer level decibel meter (like the common Radio Shack model) test that the output from Tools > Advanced Tools > Audio Calibration... > Volume calibration produces level readings that match between channels, including the subwoofer?

I want to document this model and make it the recommended approach for level calibration (unless you use convolution and a microphone), but I want to be certain we have everything correct (+10dB for subwoofer, reasonable frequency range for the narrow-band noise, etc.)

Thanks for any help.

I have the requested Radio Shack shack SPL meter.  As far as the system goes, in my home theater I've got a Denon AVR-988, 3x AV123 ELT525Ts across the front, Emotiva ERD-1s for the side and rear surrounds, and a Boston Acoustics PV600 for the sub.  I've used the Denon's Audyssey MultEQ to set up the system as well as running manually through MC's audio calibration.  Does my system qualify as well calibrated for your needs?

-John

[Matt]

I have the requested Radio Shack shack SPL meter.  As far as the system goes, in my home theater I've got a Denon AVR-988, 3x AV123 ELT525Ts across the front, Emotiva ERD-1s for the side and rear surrounds, and a Boston Acoustics PV600 for the sub.  I've used the Denon's Audyssey MultEQ to set up the system as well as running manually through MC's audio calibration.  Does my system qualify as well calibrated for your needs?

Sure.  Play the calibration files through each channel.  Adjust the volume slider (or amplifier volume) to get to 83dB on the meter on the left channel.  Then play all the other channels and see if they match.

If everything matches, we're good. 

If it doesn't, there's a problem in our test clips, the meter, or your calibration.

Thanks.

[skeeterfood]

OK, I'll try it out tonight and let you know how it goes.

-John

[mojave]

I used the volume calibration in JRiver last weekend for my 7.1 system with a Galaxy CM-130 SPL meter. It seemed to work very well, but I didn't test the subwoofers. I was planning to do some more calibration and EQ before doing a level test and I never got around to it. My system is just the HTPC to the audio device to my amps.

Instead of creating files, I just hit Play. Does this still go through the DSP stack? It didn't seem like it was, but I didn't check into it besides just the initial thought.

John, my parents are enjoying my former ELT525T's and I'm enjoying my LS-6's.

[Matt]

Instead of creating files, I just hit Play. Does this still go through the DSP stack? It didn't seem like it was, but I didn't check into it besides just the initial thought.

Yes.  There's no need to create files with the new system -- the test clips just play like any other file but are generated on-the-fly.

I'm particularly interested in the subwoofer level, since that's the trickiest.  If you could check if the test tone gets close to what a more sophisticated calibration provides, I'd appreciate it.

[bobkatz]

Yes.  There's no need to create files with the new system -- the test clips just play like any other file but are generated on-the-fly.

I'm particularly interested in the subwoofer level, since that's the trickiest.  If you could check if the test tone gets close to what a more sophisticated calibration provides, I'd appreciate it.

Matt, there's a basic endemic problem with your calibration tones! They come out of all speakers and alll drivers at all times. Your first goal is to give the user a per-channel choice of output. You need to send the tone to one channel at a time.

Most people do not have a way of muting their individual loudspeakers easily and besides, for bass managed systems, the subwoofers are going to get all the information from all the low pass of all the incoming multichannels so they will be too hot because the user probably cannot mute the incoming signal prior to the bass management filter, if you get my drift. So your first set of tones should be -20 dBFS wide band pink noise, sent to L, R, C, LFE, etc. one at a time.

Your next set of tones is uncorrelated wideband pink noise so the user can see that the signal goes up about 3 dB as he doubles it, for example, measure left channel, add right, it goes up 3 dB. If you generate uncorrelated pink noise and give them check boxes to turn on and turn off individual channels they can check that.

Your next set of tones is correlated wideband pink noise so the user can confirm the polarity, frequency-response match, and relative delay of the loudspeakers. In a perfect room, with the mike exactly in the center, if he turns on left, and then adds right to it, level should go up about 6 dB, for example.

Then do similar with the narrow band pink noise.

You're also offering multiple sample rates. For the wide band pink noise, regardless of the sample rate, you should filter your pink noise low pass to 20 kHz so all sample rates measure the same level.

Thanks to you and Mojave, I "cracked the code" in the parametric equalizer and I'm in the middle of resetting up and recalibrating my 5.1 system for use with JRiver playback. I'll be so glad to sell that Marantz A/V preamp, though it works beautifully, I think it's much cooler to play Blu-Rays from the computer instead. Your Wiki covers all of that very well. Had to buy the DTS HD decoder from Arcsoft, it seems.

[Matt]

Matt, there's a basic endemic problem with your calibration tones! They come out of all speakers and alll drivers at all times.

They should not.

Double check that you're using:
Tools > Advanced Tools > Audio Calibration... > Volume calibration

Also, pick the channel format that exactly matches your output so there's no up or down mixing.

[mojave]

Matt, there's a basic endemic problem with your calibration tones! They come out of all speakers and all drivers at all times. Your first goal is to give the user a per-channel choice of output. You need to send the tone to one channel at a time.

It is one speaker at a time for me using the settings Matt mentioned.

Had to buy the DTS HD decoder from Arcsoft, it seems.

I downloaded and installed the free trial version of Arcsoft. That will help for a while and you may find you can't delete the entire trial from your system - especially if a file is in use.

[bobkatz]

It is one speaker at a time for me using the settings Matt mentioned.

The only choices in the submenu for playing that I see are 5.1 and 7.1 and stereo under channels. How can you isolate just the left front, for example? Or am I supposed to use the "make files" button that's in there?

[bobkatz]

I downloaded and installed the free trial version of Arcsoft. That will help for a while and you may find you can't delete the entire trial from your system - especially if a file is in use.

Anyway, it's too late for a nudge nudge wink wink. I already bought Arcsoft :-(.

[dean70]

Has this function changed recently?

I was not able to get a 83db reference level on an SPL meter with the Sub channel when I went through the level cal about 4 weeks ago. Ended up setting main levels, then used move sub to left channel function in dsp studio and compared that result with the actual Sub playing to get the matching level.

[bobkatz]

The only choices in the submenu for playing that I see are 5.1 and 7.1 and stereo under channels. How can you isolate just the left front, for example? Or am I supposed to use the "make files" button that's in there?

Hmm, I'm beginning to wonder if my special routing and high pass/low pass tricks in the parametric are interfering with Matt's playback? But anyway, I can't see how you can have individual channels, one channel at a time selected without a GUI interface for selecting them...

[Matt]

But anyway, I can't see how you can have individual channels, one channel at a time selected without a GUI interface for selecting them...

When you play the 'Volume calibration' calibration file, one track for each output channel will be placed in Playing Now.

You switch between them like you would switch between any files (previous / next, double-click in Playing Now, etc.).

[bobkatz]

When you play the 'Volume calibration' calibration file, one track for each output channel will be placed in Playing Now.

You switch between them like you would switch between any files (previous / next, double-click in Playing Now, etc.).

Too simple! I'll check it out. When I hit play I got the signal out of all the speakers. I never thought to look in Playing now. Call me a "JRiver newbie". Sorry.


BK

P.S. I have to figure out the best way to check the performance of the JRiver-created crossover and routing and leveling (and EQ) with an external analyzer given that I can't get live input going. I'll try live input again as this is a new computer with a new Lynx Interface and maybe the multi-client ASIO will do the trick this time. To do this properly I'd have to have live input with six discrete input channels in order to send external test signals to each output channel.

[bobkatz]

It is one speaker at a time for me using the settings Matt mentioned.
I downloaded and installed the free trial version of Arcsoft. That will help for a while and you may find you can't delete the entire trial from your system - especially if a file is in use.

Mojave, maybe you can tell me. Arcsoft installs a startup file (TMT something) that's running in the background as seen in Task manager. If I'm not running Total Media theatre, do I need any of that stuff for JRiver to see and use the DTS HD decoder?  I did copy the DTS .dll file to the place recommended in the wiki. In other words, how much of Arcsoft can I uninstall and still get the DTS decoder?

[InflatableMouse]

Bob,

You can uninstall everything, all you need is the dtsdecoderdll.dll file. You need to copy it as per the instructions here.

In the AudioPath (DSP Popup screen) should show you whether DTS-MA is being decoded or not (when you're playing DTS-MA track, obviously ).

[mojave]

In Tools > Options > Audio you need to set Track Changes to gapped or gapless instead of crossfade. When you use crossfade, you will have two speakers playing at a time until the fade is done. I had this happen the first time I switched speakers and I thought something was wrong. However, initial playback for me has always been just the left speaker.

Matt - It might be helpful to override Track Changes and use a small gap between track changes with the volume calibration.

[bobkatz]

In Tools > Options > Audio you need to set Track Changes to gapped or gapless instead of crossfade. When you use crossfade, you will have two speakers playing at a time until the fade is done. I had this happen the first time I switched speakers and I thought something was wrong. However, initial playback for me has always been just the left speaker.

Matt - It might be helpful to override Track Changes and use a small gap between track changes with the volume calibration.

I'm puzzling my way through this. I had not seen the multiple channels options in the playing now because I started with full band pink noise, which only comes out in 5.1. Matt, can you please make all of the calibration signals available on a per speaker basis as you did with the "volume test signal". That way I can test with a true RMS meter, and so on and confirm that all your settings for all your files are as correct as possible. I know you are especially interested in the subwoofer test signal, but I like to start from first principles and then work my way down.

The other good news is I got line input working good enough (with some tolerable glitches) using ASIO multi client to be able to use REW or other analysis program to test and set up the JRiver equalizer, crossover and routing I've created for multichannel, so this weekend is going to be fun!

[bobkatz]

I spoke too soon. Setting Live input to greater than 2 channels causes a big crash situation. (I set it to 6 channels). JRiver locks up, ASIO multiclient still thinks that it's connected to JRiver, and restarting JRiver doesn't help. It becomes a big round-robin until you're forced to reboot the machine.

So I'm not sure how much line in testing I can do with only a stereo input test signal. Maybe I can get clever with rerouting in front of my regular routing in a second instance of the parametric. Probably can.

So anyway, careful about trying live input with greater than 2 channels.

It's still going to be a fun weekend, but it will be a brain-teaser as well.

BK

[mojave]

I have used Task Manager to close the Asio Multi-Client when having an issue with JRiver. I was then able to restart JRiver and use it again without a reboot. If you are running Media Network with JRiver then you need to close JRiver (actually the process) in the Task Manager, too, because closing the program still leaves the "Media Center 18.exe" process running.

[bobkatz]

I have used Task Manager to close the Asio Multi-Client when having an issue with JRiver. I was then able to restart JRiver and use it again without a reboot. If you are running Media Network with JRiver then you need to close JRiver (actually the process) in the Task Manager, too, because closing the program still leaves the "Media Center 18.exe" process running.

When I had the crash, I tried finding the multi-client in Task manager and wasn't able to find it. It might have been there, but anyway, I'm wise to the situation! Thanks,

BK

[bobkatz]

Well, I had fun and the weekend is only half over. Played two beautiful Blu-Ray movies and boy do they look and sound great in surround in JRiver! I'm very impressed by the flexibility of JRiver Media Center, which has more features and better reliability than most "professional" audio programs I use! I spent all day Friday and Saturday calibrating the system using the JRiver Parametric for all routing and crossover.

Matt, I believe that your subwoofer test signal is correct, maybe 1/2 dB low. At those low frequencies, the SPL meter jiggers quite a lot. However, my Radio Shack meter, C weighted, reads about 3.5 dB lower than my Josephson C550 with your subwoofer test signal. I believe that is due to low frequency losses of cheap test mikes, preamps, etc. at extreme low frequencies. That's the major obstacle to getting the subwoofer test signal right. I didn't look at the bandwidth of your subwoofer (LFE) test signal, but if it goes down to 30 or 40 I suggest you narrow it up with fairly steep slope and try to cover just the range from 60 or even 70 to 100. That way you stay out of the low frequency weaknesses of cheap preamps, cheap measurement microphones and SLMs. I suggest that with an FFT, measured with 1 octave smoothing, at the desired frequency (say, 70 Hz), it should read on the FFT as 10 dB below what a wideband pink noise signal reads whose RMS is -20 dBFS. (The 10 dB offset is what you've done so the untutored user will make the LFE read the same as the mains with your test signals).

I'd like to test this concept if you're game for it, let's see if we can create a subwoofer test signal that's more robust and translates better to cheap SLMs like the Radio Shack.

Measurement equipment i used: Room EQ Wizard with 500 ms. Hann window after the impulse and loopback mode to verify latency channel to channel. Since I could only test two channels at a time because of some weakness of MC's line in mode, I used the second parametric to reroute incoming channels into the main (crossover/eq) parametric. Main Microphone: Josephson C550 (1/2"). Preamp: Rane. ADC: TC Electronic. Calibrator: ACO Pacific Model 511E. Sound Level Meter: Audio Toolbox. Alternate SLM: Radio Shack with 1" capsule. Alternate mike: 1/2" capsule built into the Audio Toolbox. All three microphones and both SLMs calibrated to read 94 dB SPL with the 1 kHz calibrator in the Aco Pacific C-weighted (reads the same at 1 kHz A weighted anyway).

Ironically, I discovered the built-in mike in my professional Audio Toolbox is worse (less accurate) at low frequencies than the Radio Shack microphone since the Audio Toolbox mike reads 1.7 dB lower with full range pink noise than the C550! But the cheap Radio Shack mike, C weighted reads exactly the same as the expensive calibration microphone from Josephson! On full range pink noise, C weighted. So when it comes to judging low frequency accuracy, watch out! The only time all three microphones agreed was with 500-2 kHz narrow band pink noise. So that's the best level-calibration tone except for this issue with the LFE that I hope we can improve with the above suggestion.

Attached is a 1-octave-smoothed measurement of my left front speaker with LFE overlaid, and the same measurement 1/6 octave smoothed. (Measured with REW's sweep signal). Notice how the 1 octave analysis helps us prove the "center of gravity" of the LFE is exactly 10 dB above the "center of gravity" of the midrange.

This response is much looser than I got with Acourate Convolver. Actually, I added another parametric filter around 80 so it's even smoother than the picture attached. Anyway, it still sounds great on Blu-Ray movies and I'm so glad JRiver had the facility to implement a crossover and some correction equalization. There was no latency with your EQ, in fact I had to add 2 ms. to the audio in JRiver to get the lip sync right.

Hope this helps!

[Matt]

Thanks for the testing.

I'll look in the code on Monday and post a little more technical detail about the subwoofer test tone, and then we can discuss how it might be improved.

Cheers.

[Matt]

I'm a little slow to follow up on this, but here are the technical details for the level calibration tones:

Main speakers: Instrument grade pink noise filtered at 48 dB/octave using a low pass and high pass set to 500 and 2000 Hz, then calibrated to the reference level of -20 dBFS RMS
Subwoofer: Instrument grade pink noise filtered at 48 dB/octave using a low pass and high pass set to 25 and 150 Hz, then calibrated to the reference level of -20 dBFS RMS (and it accounts for the standard +10dB subwoofer calibration)

Feedback welcome.

[bobkatz]

I'm a little slow to follow up on this, but here are the technical details for the level calibration tones:

Main speakers: Instrument grade pink noise filtered at 48 dB/octave using a low pass and high pass set to 500 and 2000 Hz, then calibrated to the reference level of -20 dBFS RMS
Subwoofer: Instrument grade pink noise filtered at 48 dB/octave using a low pass and high pass set to 25 and 150 Hz, then calibrated to the reference level of -20 dBFS RMS (and it accounts for the standard +10dB subwoofer calibration)

Feedback welcome.

Feedback or just noise  :-).

I'm writing to give feedback, but consider it noisy feedback!

Matt, is this the same noise as what I tested a few weeks ago? If so, in your "copious free time" and only if you want to go through with it, my suggestion is to raise the high pass frequency for the subwoofer to see if it gets away from the poor ELF response of cheap microphones and preamps.

This idea is far from urgent as the whole idea of narrow band noise is not an exact science, and I personally think you have better things to do! Anyone who wants to do it right should do it right, and these alternative ideas are always going to be approximations. In the larger scheme of things, if an amateur setup person ends up with as much as 2 dB level error on his subwoofer because he chose to work with a cheap mike and a simple cheap SPL meter, that can't be too bad, can it?