Sound Frequency Actor
I am in the process of building an User Actor that outputs the energy levels and triggers that occur over the frequency range 20Hz to 20KHz when listening to a sound file. The response of the triggers seems inconsistent with the trigger level settings and energy levels. I have setup a repeatable test and collected some data.
Test 1A: movie player vol=50, frequency=125Hz, slope=50Hz, width=100Hz trig level=100%. Results from five cycles showed an average energy max of 6.5 and 13.6 triggers. Why would there be any triggers if the max average energy was 6.5 and the trig level was 100%?
Test 1B: same as 1A except the trig level was 50%. Results from five cycles showed and average energy max of 6.3 and 14.8 triggers. Why would there be any triggers if the max average energy was only 6.3 and the trig level was 50%?
Test 1C: same as 1A except the trig level was 0%. Results from five cycles showed and average energy max of 6.9 and 14.2 triggers. This is the only test where I expected the trigger count to be greater than zero because the trig level was 0%.
I reran tests 1A, 1B and 1C with the frequency width at 1500Hz to cover the entire range of 500Hz to 2000Hz. The pattern of the results was similar.
Am I misunderstanding the relationship between energy, frequency, and trig level? Is the trig level setting a percentage of the maximum energy heard? Seems like the trig count would max out when the trig level is 0% and be small when the level is 100%.
Is anyone using Sound Frequency Actors with trigger level settings and getting results that meet your expectations? Thank you.
Perhaps you can share your test file, so that I can take a look and try to understand where things are going wrong?
bonemap last edited by
Yes, sharing your file will be helpful to emulate your test. Can I also ask about how you have set up the routing ?
You are talking about the Frequency Watcher so it is not clear how the MoviePlayer volume is associated with the algorithm. Are you playing an .mp3 through a MoviePlayer and using SoundFlower or BkackHole to then get the audio into the Live Capture channel for frequency analysis? If so what is the gain setting?
My assumption is that using the frequency slope may have an impact on your trigger. I say this only because I built a macro that attempts to divide frequency ranges without overlaps. And I believe I did not have the issue you are describing with that prototype.
But I have noticed that there are inconsistencies in the frequency feedback response that appears to make it rather difficult to calibrate an accurate frequency analysis system using the Frequency Watcher or the QuickTime frequency bands of the MoviePlayer.
Here is the Izzy patch I am using. Look at the first scene. I am using the movie player to play an
.mp3 file via a powered speaker. I am using a Blue Yeti mic to listen via Live Audio capture at 0db.
To run a test: 1. set up a mic input. 2. verify live cap gain is 0db. 3. set the MBP vol out at mid level. 4. set up the movie player vol slider to 50. 5. with the green sliders, pick whatever slope, width and trig level you want to observe. 6. press GO to start live audio capture. 7. press PLAY to start the sound. 8. after about 12sec the test is complete. 9. observe the max sound level heard and the trigger count(number above the T at the green slider). 10. press RESET before running another test.
Note: During a test the number to the left of the trig count is the sound energy and those numbers seem to be consistently higher than the max sound level heard.
bonemap last edited by
There is a lot of really nice patching going on there!
Can I just bring your attention to the User Actor that is doing all of the frequency analysis. It appears to me, in the version you have shared, that you have a double set of Frequency Watchers inside the User Actor and that the input and out put are not connected to the same series of Frequency Watcher actors there. Therefore, the patch is not providing a clear indication of either the interactive calibration of the frequency bands or the measure of output.
Russell, thank you for pointing out the two sets of Sound Frequency Watchers in my User Actor. I went back and cleaned it up and revised it to rev3. While in there digging around with the patch running I kept noticing the large value differences between the Sound Level Watcher output and the energy output levels of the Sound Frequency Watchers.
Question: Is the volume output of the Sound Level Watcher measuring the same "thing" as the energy output levels of the Sound Frequency Watchers?? I don't think so.
I decided to revise my User Actor and my Izzy patch to observe the seven energy levels instead of the volume. This seems to resolve my question about trig level and trig count. I am attaching the revised Izzy patch.
@DusX Sorry to jump ahead with a revised patch before hearing your comments.