I had a conversation recently with a user testing the trial device regarding changing the default sampling rates selectable via the config software. They were keen to record at a sampling rate of 384Khz. I thought I'd therefore open a thread on the forum to allow the AudioMoth team to reply with a little more information regarding what's possible with the current device, as I believe at present 222KHz - 240KHz with short buffer recordings is the max to retain stability (and SD write speeds / cart types also need to be accounted for).
We've just updated the configuration app to allow recording at 240kHz and 320kHz. These options are disallowed by default and must be allowed using an option under the ‘Audiomoth’ menu before they can be selected. They should be used with a fast SD card - either a SanDisk Extreme or a SanDisk Extreme Plus.
320kHz would be the absolute minimum when recording UK/European bat species. For tropical bats you would need 384kHz sample rate.
Thanks a lot Alex! I will try to record at 320kHz then and I will send some feedback soon! :) Congrats for such awesome work!
We've updated the configuration app - version 1.0.4 - for 320kHz and 400kHz since both seem to work well using a SanDisk Extreme card.
Great news! Thanks Alex!
no indication of drop-outs)Hi Alex, just as a matter of interest is there some reason for going up to 400ksps instead of the industry standard 384ksps? And can you say what happens in the event of a deficiency in speed - do you just leave out a buffer full of data when it happens (and how big is that buffer?) or is the event flagged in some way? I am always suspicious of many of the bat recorders which claim that they do not have any recording 'drop-outs' but I am not sure if they measure it properly. If (for example) they record for 4 minutes - do they just write 4 minutes worth of samples successfully?
(cont) or (better) do they sample for an externally clocked 4 minutes in which case a reduced number of samples in the recording will indicate the amount of data that has been lost.
Yes, we're currently working on an improved version with a 384kHz sampling rate. In the current firmware we run the ADC in free-running mode, continuously taking samples, and we adjust the frequency of the oscillator that controls the ADC to get the correct sampling rate. We can't do this for the very highest sampling rate however.
Our new firmware continoulsy runs the processor on the fastest and most stable oscillator, and uses a timer to trigger ADC conversions at 384kHz. We then oversample in firmware to deliver 192, 96 and 48kHz. We do the same at 256kHz to deliver 32, 16 and 8kHz samping rates. The new approach gives slightly cleaner recordings and much more flexibility in terms of sample rates. It is slightly more energy intensive, but not signficantly so, and for most recording deployments the SD card capacity is the limiting factor, rather than the battery life.
At 384kHz we are running the microphone well outside of its design specification. However, it does seem to be reasonably sensitive still. The recording below shows bats at 48kHz with reasonably strong harmonics still visible at 144kHz, and we recorded lesser horseshoe bats at 115kHz. Individual microphones will generate some temperature sensitive noise in the range of 100 - 150kHz which shows up as a horizontal band on the sonogram.
https://www.dropbox.com/s/d1gle82pm28cq7j/59CD4D0D_384.pdf
These are generally easily discernable as artifacts, rather than natural signals, and are removed by oversampling in the lower sample rate recordings. The recording below is the same as above but with 2x oversampling to give an effective 192kHz sampling rate.
https://www.dropbox.com/s/ghg3p5k9z4vptyb/59CD4D0D_192.pdf
The new version requires the firmware on the device to be reflashed (along with an updated configuration app) so we'll release this in a few weeks when we've had more opportunity to test it out in some real deployments.
Thanks for that info . I used too use 192ksps (direct into the Realtek sound card of a laptop) with the MEMS microphones and found it very useful for identification at least. In my part of the UK we don't get any LHBs so no problems there. However, using my Petterson M500-384 and an EM3+ (the former an FG the latter a MEMS) I find Pips of both flavours, as well as several Myotis extending to 140kHz or so. Not of importance for ID purposes which is all that many people are interested in, but significant for those of us interested in the bioacoustics of bat calls. Presumably the 384kHz without oversampling is restricted to 12bits (or less in terms of ENOB) - but still better than the 40dB dynamic range we had to put up with recording to Mag tape.
Yes, 384kHz is 12 bits and then we get another bit for each halving of the sample rate; up to the full 16 bits at 32kHz, 16kHz and 8kHz.
Hello. I am testing my Audiomoth device in Southern Chile, recording dusk-night-down period for a week in a rural area within coastal Biobío Region. The sampling rate I used was 48 kHz. I found some very intriguing sounds wich rise from sonic to ultrasonic spectrum, and I need to be sure that they aren't produced by the recorder itself. Someone there can offer some help? Below there is a screenshot from Adobe Audition spectrogram of one of the files. Thank you!
Hi, I don't know what they are but they are definately real, not an artifact of the recorder. Can you post a recording somewhere - SoundCloud - so others can try to identify it.
Hi Alex, I´m from the University of Greifswald and I will do my PhD thesis on lesser horseshoe bats. I will record some roosts during summer using Audiomoth. Therefore the information you provided on Dropbox above would be really helpful. Could you upload them again? Thanks for your time.
Available again here:
https://www.dropbox.com/s/rmtr60w2xbizq3w/59CD4D0D_192.pdf?dl=0
https://www.dropbox.com/s/glax574mamo860j/59CD4D0D_384.pdf?dl=0
The actual recording is here:
https://www.dropbox.com/s/1aw9k4g7vtzsbxw/59CD4D0D.WAV?dl=0
Thanks a lot.
Hi! I am from Osmania University, working on bats of Central and Western India, I wanted to know that can i select the range of recording to a specific frequency? like if i want to record only between 16kHz to 60kHz where i can get specific Vespertilionidae and Emballonuridae calls?
There is no filtering on the actual device by default but you can record with a suitably high sample rate and apply the filtering that you need afterwards to the resulting recording. If you avens specific need to filter on the device (so you don’t record people) we can implement a custom version of the firmware to do that.
@Alex Rogers thank you! i am eager to explore the custom firmware.
@Deepak Tarun Hi, What's the end application? Why do you need the high pass filter?
Hello, I also would like a filter to limit recordings to bats. This option is standard in recorders for bats and greatly reduces the number of recordings to a few gb during a multiple night sampling interval. Even with the sleep for 5sec and record for 5sec options, which means data are never recorded, the cards fill up, which means data are never recorded and also take a long time to process. Ultimately, this means the audiomoths can't compete with commercial devices that capture the precise types of audio that we desire versus simply recording everything and making us try to get the calls we want from that enormous data swath. We don't have time in the back end to be worth the reduced cost.
Hi, We have a firmware with triggering for bats that we plan to release in a month or so. A few people have been trialing it for us over the last couple of months. We've been delaying due to the complexity of supporting multiple versions of the configuration app but we have a nice solution to that problem now that we're in the process of implementing. Alex
@Alex Rogers Perfect! That's really what we needed. Thank you for the prompt reply!
@Alex Rogers Jumping late in the discussion. If you need another person to test recording tropical bats and comparing with SM4BAT FS, let me know.
If you need another tested, let me know.
Same here! @Alex Rogers If you need more tests in the Mediterranean/Pyrenees area, please let me know :)