Measurements to know if my field-recordings are good?

Question

Fieldwork means having all the equipment in place to record the animals (in my case bats) when I see them. I try to look at the audio as soon as they're recorded to check if the settings are okay. I can't check every single recording and optimise settings in real time as this would be too cumbersome , and my presence may interfere with the animal's activity.

How do I know my audio is 'good enough'. I only do the basic checks of 'Is there clipping/saturation?' and 'Is the signal visible on the waveform'.

I tried looking for resources that have a checklist of 'things to look out for' but can't find any. This knowledge seems to be very word-of-mouth. Would anyone mind sharing their own 'field-recording checklists' or any workflows out there?

Answer 1

In addition to:

• (is there clipping/saturation)?
• (is the signal visible on the waveform?)

You could also ask:

• what is the SNR (signal-to-noise ratio) of the signal?
• if the signal is too faint, what are my gain options (+6, +12, +30, etc) available and which would be most appropriate for the next recordings?
• if multiple instruments are deployed simultaneously and near one another (or together on one soundcard), is there any coherence in signals received?
• do tap tests to sensors at the start and end of deployment, and then calculate the clock drift (is this what you expect it to be)?
• create an LTSA (long term spectral average) plot, or even just a spectrogram of a brief instance, and check if there is any electronic noise (e.g. perhaps there is a 60 Hz hum or something, and better grounding of instruments can help alleviate this). This noise can usually be filtered out, but trouble arises if this overlaps with your signal of interest.
• create a PSD (power spectral density) plot of a few files to see how much unexpected noise there is at frequencies both overlapping and not-overlapping to your signal. This can help to identify electronic noise (as above) or to identify a sound in your environment that you were unaware of, but might influence the presence of the animal you are trying to detect.
• calculate the background noise of a given band or the estimated source level of a given signal, and see if these are within ballpark range of being reasonable. If not, it could tell you that your calibration value needs to be scrutinized.

I'm sure there's more - hope this helps