Staacks
commented
5 months ago - Let's start with the meaning of "a.u.". This stands for "arbitrary units", indicating that these are relative units as absolute values with a "proper" unit are not available. In case of audio, the microphone is not calibrated such that we can get the sound pressure in Pascal, but instead we get values that are normalized to a device-specific maximum of 1. Hence, no unit can be given, which is commonly indicated as "a.u.". (You are not the first to ask this, so we should probably check if enough space is available to replace this with "arb. unit" instead.
- I think I see what you expected how to use the acoustic stopwatch, but it is used a bit differently as you might expect, mostly because of the way that everything in phyphox is set up in the same way through a common file format. The button in the top right corner (the "play" triangle) start the measurement regardless of which tab you have open. There is no need to move to the many tab unless you want to measure multiple events, but you can also switch between them during a measurement as you like, it's all the same measurement and the first tab only shows the first time interval as a simplified interface for students. I would suggest, that you try the stopwatch on the first tab first: Press the triangle and then press reset if the counter is not at zero. Clap your hands to start the counter and clap a second time to stop it again. If it starts before you clap or if it stops immediately with the first clap, you need to increase the threshold as it is too sensitive. If nothing happens when clapping, it is not sensitive enough and you need to lower the threshold. As said, the units are arbitrary on a scale from zero to one and the ideal level depends on the sensitivity of your phone. From there you can check the other tabs, which simply allow to measure subsequent claps in different ways.
- Your example of a fan is way too fast for the acoustic stopwatch and I also think that the sound is too uniform to act as a trigger signal for the stopwatch. You should not try to measure an interval between changes in amplitude but instead analyze for periodicity. Testing Audio Correlation and/or Audio Spectrum was the right idea.
- Audio Spectrum calculates a Fourier transform of the sound data and has some mathematical limitations. In particular, the resolution of the spectrum depends on the length of the time period that is analyzed. In case of your phone with a fix sample rate, this length and therefore the resolution of the spectrum depends on the number of data points that are recorded and analyzed. For most applications this is a compromise between "How fast can I detect changes in the spectrum" and "How precise can I determine a frequency". In your case I would assume that you do not have to be fast, so simply increase the number of data points to the maximum. To do so, switch to the settings tab and tab the 32768 button. Your spectrum should have a significantly higher resolution. If you need more, this can be increased with an external configuration, but that can quickly get out of hand as halfing the distance between data points means doubling the number of points to record. And audio autocorrelation might indeed be the better choice...
- Without seeing or hearing the exact data you are working with, I would think that the problem you are having here is that 50Hz is too low for the way audio autocorrelation is set up in phyphox. It only calculates 20ms of autocorrelation which means that it can only hold exactly one period of a 50Hz signal. I have created a version that double the recording and autocorrelation period and should therefore work with frequencies down to about 25Hz. Either open the link/file at the bottom on your phone or scan the QR code from within phyphox (the "+" button on the main screen). Be aware though that we introduced that 20ms limitation because the computation is rather intesive. But then again, we made that decision many years ago, so modern phones might handle this nicely.
audio_autocorrelation-lower.phyphox.zip
I will now close this issue. I have put renaming of "a.u." to "arb. unit" onto my todo list, but the other things are limitations of the underlying math or the modular design of phyphox. By that I mean the ability to load a new configuration with a QR code means that all experiments are designed in a similar way with the play button starting the currently loaded configuration and then having buttons in the interface that has been loaded. I see how this can be unintuitive at times, but I do not see a solution with the resources that we have available.
Closing this issue does not mean that this discussion ends here - just that it does not pop up as a coding todo item. Feel free to ask here if you have further questions or problems.
I do not understand this. The settings want a measurement in "astronomical units"(a.u. -called correlation on audio correlation page) and I lower minimum delay low. I go to final tab (many) press start experiment, and see nothing. I press to stop. I go to first tab and it seems to have caught something.
Sound source is a small desktop fan (5.5" grille) that I have teetering and when it is running it makes continuous loud tapping sounds, maybe 50-70hz as a guess. I tried "audio spectrum" but its increments are too great at low frequencies. The "audio correlation" page seems to be locking to different components of the sound (maybe you can list the different frequency magnitudes it has noticed so you can choose which to observe?) so it is no hint for what "a.u" to put in.