Synchronizing Recordings from Multiple AudioMoth Devices
I need to synchronize the recordings of several AudioMoth devices with microsecond precision. I attempted to do this by turning the devices on approximately simultaneously in default mode and recording a reference signal both at the beginning (within a few seconds) and at the end of the recording, just before shutting the devices down.
Unfortunately, the signal at the beginning and the end of the recording is sometimes not captured.
I have three questions regarding this:
How long does it take for an AudioMoth device to start recording after being powered on (in seconds or milliseconds)?
At what intervals are recordings saved to memory (e.g., every minute, every 15 minutes)?
Are any recordings lost or skipped during the saving process, or is all data continuously stored?
Viele Grüße, Elmar
Hi Elmar, Depending on the contents of the SD card it can take up to five seconds to open the file and start recording. In normal use, the AudioMoth firmware tracks this duration and adjusts it's preparation time so that the recording always starts at the required time. If you set the clock on the AudioMoth immediately before recording, and use a scheduled start time, then all the AudioMoth recordings will start at the required time. The realtime -clock drift can be up to +/- 80ms per hour so you will need to allow for that uncertainty. If you are recording outside you can use the GPS board and the AudioMoth-GPS-Sync firmware. This runs the GPS continuously during recordings and provides software to automatically synchronised recordings to +/- 1us. https://github.com/OpenAcousticDevices/Application-Notes/blob/master/Measuring_the_Accuracy_of_AudioMoth_GPS_Synchronisation/Measuring_the_Accuracy_of_AudioMoth_GPS_Synchronisation.pdf This firmware uses the pulse-per-second (PPS) output of the GPS throughout the recording. You probably won't be quite as accurate if you only synchronise the recordings at the start and end as the internal 48 MHz clock will drift slightly over the course of the recording due to the temperature fluctuations. If the temperature is consistent you'll get the best results. Also, 1 us is a lot less than the inter-sample period (22 us at 48 kHz), so you will need to be quite careful interpolating the recordings when you try to correlate the recording with the reference signal. We've done this in the past by interpolating the zero crossing times of the waveform.
The recording schedule is determined by whatever you set in the Config App. Data is written to the SD card continuously in 32 KB chunks during the recording. At 48 kHz this corresponds to 341 ms. Alex