Help diagnosing corrupt files

We just encountered a handful of files from the the beginning of a deployment that seem to be corrupt. There were many working normal files recorded before these so it seems strange. but then there are tons of files that are normal afterwards as well. I have attached a few examples here in the zip. The files are consecutive and seem to be limited to 1 day, so not a big data loss but curious if anyone can help diagnose what happened?

them here (I think...).

5 comments

5 Comments

Abram Fleishman

Apr 25

Hi Alex, thanks for your reply.

Here's the summary csv with those problem files toward the top.

Firmware was AudioMoth-Firmware-Basic (1.8.1)

Alex Rogers

Apr 25

Replying to

It's strange that these files were affected but then files afterwards seem fine. What sort of batteries are these? They don't show any voltage reduction with use until they are finally depleted when the zero length files are generated. Also what make and model of SD card is this? I'll make a Python script that will put the header back on these files. You'll be able to call it with the filename and sample rate.

Abram Fleishman

Apr 25

Replying to

Interestingly, if you open one of the WAVs in a text editor and search for the RIFF tag it finds it on line 20,000 or something like that?! (I can get info about the cards next week, but i am not sure I will be able to provide battery info. I assume they were just normal alkaline batteries but this device was deployed by a partner.)

Alex Rogers

Apr 25

Replying to

Yes, you are right. The header is at 0x4E0000 rather than 0x0 which is very strange. It is also the header for the next file, rather than the actual file. This header is in the 20231212_182000.WAV file.

I think this must be some corruption of the file allocation table, rather than the AudioMoth failing to write the header correctly.Regarding the batteries, they don't look like alkaline batteries as the voltage is very high and doesn't decrease at all across the deployment. They are possibly 1.5V Li-Ion AA cells. These have a voltage regulator that maintains the voltage at a constant 1.5 volts until the battery is depleted and then switches them off.

Alex Rogers

Apr 25

These seem to have suffered from an SD card write error when the AudioMoth tried to add the header onto the start of the file having completed the recording. The file is the correct length and the data is probably intact. I can write you a script that will add a suitable dummy header back onto these files. Can you use the 'Summarise AudioMoth Files' option in the most recent version of the Config App to generate a summary of one of the SD cards where this occurs? It will show the temperature, battery, and header information on each file in a CSV table, and you should be able to easily see all the occurrences. What version of firmware were the AudioMoth running?

Edited