Originally Posted by 3ll3d00d
and there was me thinking I have to sign off all test signals, what is the world coming to?!
since you asked nicely then I'll offer a couple of observations...
The signal is described as "Each impulse has content from 1-80Hz with a big peak at 30Hz and full band content below 10Hz with a peak at 6Hz". The terminology is unclear (use of "impulse" and "full band" to describe band limited content when I'd consider both of those to refer to genuinely full bandwidth signals) but if I were constructing such a signal then I'd guess it would be something like:
* band limited white noise at some background level
* 30Hz sine wave
* 6Hz sine wave
However the signal itself appears to be 100ms long. A single cycle at 6Hz is ~167ms so not sure how you can have <10Hz content in there. FFT analysis looks like a big 30Hz peak with some side lobes.
How was the signal created?
There is a technique one can use for low frequency alignment which lets you look at the evolution of that alignment over time and hence lets you examine alignment at the start of the wave vs alignment in the steady state condition. If you do this, you'll find that the correct delay differs according to whether you choose to align the leading edge or the steady state. Invariably one aligns to the steady state here though whether I could honestly tell the difference, I have no idea
I also have no idea whether any human can subjectively perceive a difference at this level of resolution when comparing audible content to tactile content but, if you can, it raises the Q of whether you should use a longer signal for that subjective alignment process.
Do you find you get consistent results if you blind test yourself at dialling it in? what about blind testing to pick the difference? might be hard to do the switching quickly enough using offboard dsp but would be interesting to see the results if it is possible. I can't say I'd be bothered enough to test that myself mind you
Matt.....Thanks for weighing in and doing a deep dive on the signal. When I received the file this weekend, I hastily ran it through your BEQd app to see the frequency content before testing it out since I had no idea the origin of the signal or description of what the signal was supposed to be other than a time alignment signal.
The 30Hz band was prominent along with content below 10Hz centered at 6Hz. Although that content was ~30dBFS lower in level. After your comments and looking at the waveform, I agree about the sub 10Hz content. Looks like noise down there that was being analyzed and reported since the signal is only 100 ms in length.
Interesting thoughts about leading edge vs. steady state. For the intended purposes of this test signal and using the good 'ole buttometer for measurement, I always presumed it was a leading edge phenomena we are "feeling" for especially with a signal of such short duration. When I think of steady state, I usually think of something happening for at least a second before our bodies can "feel" and get a sense of the steady state response.
If I'm understanding correctly, steady state alignment could also be misleading as it would just be a phase shift phenomena rather than a leading edge phenomena if the signal has strong sinusoidal content.
That's the way I've always thought about time aligning tactile vs. audible sounds anyway.....getting the leading edges to time align instead of steady state.
In this case, after your feedback, I now think the original intent was to use the 30Hz burst signal to energize both tactile and audible devices in the room and time aligning the leading edges of both using the simple buttometer. Although after thinking about that some more, with the signal only being 100ms in length, are we really time aligning the leading edges with our butt or just a weighted response to what our bodies feel and hear for that short 100ms burst.
Having said all that, if we use this 100ms 30Hz burst signal for time alignment, are there any pitfalls with this approach? I'm more of a structural response under dynamic load kind of guy with limited experience in the time domain and alignment of that response with audible signals which is why I'm asking.