Level matching

[Pete Schumacher ®]

In the discussion pertaining to the Iowa DIY, Dave Ellis reminded me of an issue that I'd like to get some feedback on.

In setting the signal level for each system, a test signal using Brown-Noise was used. Brown Noise is heavily weighted to the low end of the scale by definition. That presents a problem.

When speakers with deeper extension play back the signal, they'll put more low frequency energy into the room, exciting room modes as well. Compared to smaller systems with little bass below 60Hz, the larger ones will appear louder since they respond to more of the spectrum. By then adjusting the smaller one up to reach the same dB level, it will have more midrange and higher SPL compared to the larger system on material with little bass, like acoustic and vocals.

I'm suggesting band limiting the brown noise to 300Hz and up. That takes room modes out of the equation and will allow the larger systems to play at the same volume on music that has very little bass content, especially vocals.

We all know how even fractional dB differences in the overall SPL will favor the "louder" system and by high-pass filtering the test signal, you allow a more even playing field.

Thoughts?

[Jeff B.]

> In the discussion pertaining to the Iowa
> DIY, Dave Ellis reminded me of an issue that
> I'd like to get some feedback on.

> In setting the signal level for each system,
> a test signal using Brown-Noise was used.
> Brown Noise is heavily weighted to the low
> end of the scale by definition. That
> presents a problem.

> When speakers with deeper extension play
> back the signal, they'll put more low
> frequency energy into the room, exciting
> room modes as well. Compared to smaller
> systems with little bass below 60Hz, the
> larger ones will appear louder since they
> respond to more of the spectrum. By then
> adjusting the smaller one up to reach the
> same dB level, it will have more midrange
> and higher SPL compared to the larger system
> on material with little bass, like acoustic
> and vocals.

> I'm suggesting band limiting the brown noise
> to 300Hz and up. That takes room modes out
> of the equation and will allow the larger
> systems to play at the same volume on music
> that has very little bass content,
> especially vocals.

> We all know how even fractional dB
> differences in the overall SPL will favor
> the "louder" system and by
> high-pass filtering the test signal, you
> allow a more even playing field.

> Thoughts?

Why brown rather than pink?

[Pete Schumacher ®]

> Why brown rather than pink?

Wasn't my choice.

I'd use pink, but I think it tends to hurt some folks ears. But even with pink, you still need to HP filter it to take room modes out of the equation.

[DougP]

> Why brown rather than pink?

Last year brown noise was choosen because it was much easier on the ears than any other test signal we could come up with. It will be ajusted this year for the reasons Pete mentioned.

Doug

[Jeff B.]

> Wasn't my choice.

> I'd use pink, but I think it tends to hurt
> some folks ears. But even with pink, you
> still need to HP filter it to take room
> modes out of the equation.

OK. Then yes, I would agree that limiting the bandwidth of the noise for set-up purposes should help to find a matching level for all speakers.

[Jeff B.]

> Last year brown noise was choosen because it
> was much easier on the ears than any other
> test signal we could come up with. It will
> be ajusted this year for the reasons Pete
> mentioned.

> Doug

How loud do you guys set up the speakers?

[DougP]

> How loud do you guys set up the speakers?

I believe that they were set at about 80db using a rat shack SPL meter. That reading was taken in the middle of the seating area about 12 to 15 feet back from the speakers. We decided on the level by general consensus. This was nine months ago so I may be remembering it wrong.
Doug

[AJ]

If you guys have a ratshack meter, why not just run a couple of test tones? Like 1khz, 500hz and 5khz?

[curt_c]

> If you guys have a ratshack meter, why not
> just run a couple of test tones? Like 1khz,
> 500hz and 5khz?

It really needs to be an average of the response for the reasons Pete mentions. A speaker with a dip a the test tone frequencies would sound louder, and subjectively better than one that measured flat. OTOH, if it was outside of the bandwidth limited level measurements, a design with a low bass bump could sound subjectively better than one that measured flat, all else being equal.

Like everthing in speaker design, there are few easy answers...

C

[JRT]

> How loud do you guys set up the speakers?

See 4.1.4 in the whitepaper at the link:
<A HREF="http://www.dolby.com/assets/pdf/tech...sic_Mixing.pdf">http://www.dolby.com/assets/pdf/tech...sic_Mixing.pdf</A>

While that is primarily targeting 5.1-Channel Music Production Guidelines, the suggestions can also be applied to 2.0 or 2.1 setups.

In that Dolby is suggesting 85dBC uncorrelated sum reference level at the listening position, corresponding to pink noise reference signal -20dBFS (16bit PCM).

For mastering music on a stereo setup, Bob Katz at digido.com suggests 83dBC from each of a stereo pair, (86dBC uncorrelated sum) reference level at the listening position, corresponding to pink noise reference signal -20dBFS (16bit PCM). ..essentially similar.
<A HREF="http://www.digido.com/bob-katz/level...-k-system.html">http://www.digido.com/bob-katz/level...-k-system.html</A>

Here is a -20 dBFS RMS pink noise 44.1 kHz Uncorrelated WAV file that Bob provides at his website:
<A HREF="http://www.digido.com/remository/gen...-wav-file.html">http://www.digido.com/remository/gen...-wav-file.html</A>

[JRT]

For an eye opener, work the numbers on the combination of loudspeaker sensitivity and associated amplifier requirements, to able to hit 0_DBFS in the digital domain without clipping in the analog domain, while listening at mastering studio reference levels.