Room Eq'ing with RTA and pink noise

[Colvic]

Does anyone have any thoughts on room EQ'ing with an RTA and pink noise.

Firstly generating a flat response on the RTA usually = a flat sound with a particularly muted bottom end.

 

My main question is, should I RTA a room using A weighting or no weighting. Using A weighting will maintain a good bottom end to the sound since it compensates for the ears comparitive insensitivity to the low and high ends of the audio spectrum.

But, in doing this, could I be adding energy into the system where I shouldnt be.

 

Are there any good reasons why Pink noising a room with the analyser set to A weighting is a particularly bad idea?

 

If I should be analysing with no weighting, I assume I should be allowing for a non flat response at the bottom end so it doesnt sound too unnatural.

 

Does anyone have a graph of a normal sounding frequency response I can use for reference.

 

Thanks

C

[Mr.Si]

A quick google on the words "Equal loudness curve" came up with a few pages, I shall post a link to THIS one.

 

It's an interesting question you ask, and one that I'm not gonna try and answer (because my experience is less than others on this forum).

 

But A weighting is the closest measurement weighting we acousticians have to the human ear's response. (It is almost a mirror image on the horizontal plane to the weay a human's ear responds to different frequencies).

Therefore since humans will be listening to the sound in the room, rather than robots or grass, then it might be an idea to at least try it out and see if it gives much difference in result.

 

To be honest I didn't know that any weighting was available for RTA's. - which unit are you using?

 

 

one point to make:

 

is that if you only use one microhpone position, then you're only gonna be measuring the room's response at that particular position, so you should do at least 6 positions in the room, and then average the result and use that average. Otherwise you'd be prone to room modes.

 

Alternatively, you could measure over a period of time and move the microphone around in the room in a diagonal arch type of movement and this will then create an average too.

only problem here is that not many machines that I know that are used as outboard for Sound Reinforcement can take averages over time.

 

Hope this in some way helps (?!)

[misterbassman]

Smaart allows you to save snapshots of the analysis in different colours and overlay them on top of each other, this is really handy for taking multiple measurements

[griffter]

It is better to use a Transfer function to EQ a room rather than just an RTA. An RTA will only show you the frequency response of your source material + the PA in the room, Therefore you cannot tell the two apart. A proper transfer function takes a reference signal from the output of your mixing desk and compares it with your analysis microphone in the room, thus you can see just the frequency response of the PA in the room. You can then use a track or pink noise to EQ the room as the process is only looking at the difference between the two signals.

 

Transfer functions are used in software programs such as SMAART, Spectrafoo, MacFOH and Meyers SIMM analyser. These types of analysers will also have a S/N Ratio and Phase display which also help greatly in properly lining up a PA System

 

Check out http://www.meyersound.com/products/integra...ols/sim3/02.htm for some more info!

[Colvic]

To be honest I didn't know that any weighting was available for RTA's. - which unit are you using?

 

 

 

<{POST_SNAPBACK}>

 

 

I'm using a Phonic PAA3.

 

It will calculate averages when taking multiple readings.

 

Thanks for your info on Equal Loudness Curves. I suppose would be fair to assume, or at least experiment with the concept that a perception of equal loudness represented in an eq curve for a room would be more musical than setting up a flat response.

 

Indeed I've already tried this in my studio room using an A weighted RTA reading and its much nicer. Of course thats fine for playback of music but could be a disaster in terms of gain before feedback when setting up a speech re-inforcement or band PA system if we are boosting a significant portion of the bottom end.

 

I'm beginning to get the impression that using the RTA for room EQ'ing with pink noise should only go as far as reigning in obvious peaks and troughs rather than eq'ing for a flat response.

 

This assumes were using good quality equipment correctly set up in a resonable acoustic space.

 

Interestingly, as you suggest, flipping the equal loudness curve on the horizontal axis is pretty much emulating the A weighting filter in the PAA3. Apparently the increased sensitivity to the lower frequencies at higher volume pretty much follows the C weighting filters characteristics.

 

Chris

[Colvic]

It is better to use a Transfer function to EQ a room rather than just an RTA.  An RTA will only show you the frequency response of your source material + the PA in the room, Therefore you cannot tell the two apart.

 

Thanks for your response

 

I see what you are saying. But surely if I'm using Pink noise from a tone generator (having a known frequency response) fed directly to the amps, then the RTA's reading will be showing purely the effects of the room acoustic plus the PA's frequency response on the pink noise which is what I'm correcting for.

 

I can see that using a transfer function is a more flexible method as you dont have to use pink noise but isnt the pink noise method still an accurate one?

 

Or am I missing something crucial.

 

Im using Phonic's PAA3 BTW.

 

Chris

[Mr.Si]

but isnt the pink noise method still an accurate one?

 

Or am I missing something crucial.

 

<{POST_SNAPBACK}>

 

Pink noise is equal amounts of energy at every octave band frequency, rather than white noise which gives out more energy the higher the frequency.

 

not sure what Brown Noise is, though I guess it's not pleasant either.

 

Whereas using any type of CD music would be more inaccurate still.

[dbuckley]

My main question is, should I RTA a room using A weighting or no weighting.

<{POST_SNAPBACK}>

 

In my opinion, you should use no weighting when equalizing.

 

The idea of weighting is for quoting a single number for SPL purposes when the input audio may be pink noise, a single sine wave, or music, or the sound of a generator running. An RTA measures each band (usually of a third octave width) independently, so fudging for ear response versus numbers is not appropriate.

 

However...

 

a) I dont like the sound of a flat system, it honks a bit too much for me, so somewhere around 2K-4K will be dropped a little...

 

b) RTA is now recognized as being inadequate, you need a transfer function machine to do the job "properly", ie smaartlive, however, if you have these tools you need either the first principles understanding or training to use them..So I'm still using RTA :-)

[Mr.Si]

a) I dont like the sound of a flat system, it honks a bit too much for me, so somewhere around 2K-4K will be dropped a little...

 

 

<{POST_SNAPBACK}>

 

That's due to one's ears being most sensitive at between 2 - 5kHz and thus those freq's sound loudest to us all. - hence the cutting of these frequencies being carried out.

[Simon Lewis]

Although it may be a matter of semantics, you cannot "equalise a room". The founder of the phrase now says he regrets ever uttering it....

 

The closest you can get to 'equalising a room' is probably by using a JCB. The point being that the acoustics will stay the same, no matter how much we twiddle our equaliser knobs or faders.

 

Useful though RTAs can be, they do not allow you to separate the direct sound of the PA from the reflected sound from the room. Therefore the spectral response given by an RTA is often incorrect. Furthermore, 1/3rd octave resolution does not allow you to recognise comb filtering, and you cannot measure the phase response of the system.

 

Smaart, TEF, SIM etc. allow some degree of time windowing, which allows the user to see more of the loudspeaker response than the room response.

 

There are some stripped down versions of programmes like WinMLS that offer quite good functionality, if you want to explore this further...

 

It's not usually appropriate to use A weighting for any measurement or equalisation, as most events are at a much higher SPL than this equates to.

[griffter]

The main difference is that the transfer function contains both time and frequency information for the system under test. The Phase and Coherence (S/N) readouts give further clues as to what can and can't be solved by EQ. For instance if you see a dip in the coherence graph coupled with a bump/dip in the frequency response this could be a reflection, which would only be solved by acoustic treatment or adjusting the placement of speakers. It wouldn't be possible to see this on a simple RTA analyser.