Print

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

[mattkhan]

it looks like it is high pass filtered, this is with full bandwidth pink noise and no signal processing inside jriver

in fact the data below 20Hz seems pretty much like junk, e.g. if I just play a low frequency sine wave in then I get the same straight line rolloff

[RD James]

Try Voxengo Span http://www.voxengo.com/product/span/

[mojave]

Span does the same thing. Even increasing the block/bin size didn't change anything.



I use SMAART for  transfer function comparisons. It's resolution is to 11 Hz, but it is complete flat to that. This is with the pink noise being generated by JRiver and routed through SMAART digitally using my MOTU drivers.



When banding is turned off in SMAART, one can see the raw FFT. This is the same as generated by Spectrum Lab and is incorrectly used for "frequency response" in the Data-Bass analysis of bass in movies.

[mojave]

By the way, mattkhan, SMAART does transfer responses with live audio. For example, I can route the subwoofer output directly to SMAART in the MOTO mixer and then route the microphone feed to SMAART and it will compare the two showing the difference on the screen. Its pretty cool. However, it shows that the bass in the room is always louder than on disc. I'm still investigating.

Here is how the JRiver pink noise looks routed back through REW.

[mattkhan]

I repeated your test in REW itself using its RTA and saw the output was flat to 10Hz, I thought its full range pink noise was now flat all the way down but perhaps it is still 10Hz and white noise is all the way down. Anyway white noise looked as expected in its RTA so this confirms this is a defect in the analyser view. It would be nice to fix this as the analyser is a nice and easy way to get a quick idea of the content being played.

However, it shows that the bass in the room is always louder than on disc.

what do you mean by louder in room exactly?

[Matt]

I'm thinking about increasing the filter size that the analyzer uses to get even better at super low frequencies.

I sent the team an email and we'll see what they say.

[Matt]

Coming next build we'll try this:
Changed: Increased the resolution of the Analyzer so it should be much better at really low frequencies.

It seems pretty good to me, but of course testing appreciated.

[mattkhan]

thanks, I will test it out when it is available

[blgentry]

I repeated your test in REW itself using its RTA and saw the output was flat to 10Hz, I thought its full range pink noise was now flat all the way down but perhaps it is still 10Hz and white noise is all the way down.

It's worth saying here that pink and white noise are statistical constructs.  Meaning that any given frequency will be present *at some point* in the signal, but not necessarily at any given time.  Which is why pink and white noise "bounce all over the screen" when viewed on a responsive RTA.

JL Audio has a very interesting adaptation of this that they use to calibrate their auto-eq systems:  They have a special signal that they construct, which has sine waves at 31 center frequencies, all at equal levels.  It sounds like a bad, and very full range, organ being played all at once.  But it's nice because the signal level is *dead flat*; no bouncing.

If you're really into doing these kinds of measurements at low frequencies, you might steal this idea and make Matt Khan's special low frequency organ note of death:  Something like 20 to 30 sine waves, at various low frequencies, all at equal signal levels.  Then there would be no doubt whatsoever of the content of the test signal.  One naive version of this would be 1 to 20 Hz, at 1 Hz increments.

For bonus points, what signal level in each sine wave is required to achieve right at 0 dBFS total at the output for these 20 sine waves added together?

Brian.