Hi everybody, I am new in this group, and I hope to be right here with my question.
I am working on a projekt for monitoring of birds. Therefore, I like to place several numbers (>50?) of MP3 recording systems in the observation area for a possibly long recording reriod. In the first step I like to collect the SD cards with these files from the recorders, put them into my home computer and make a sound analysis with "BirdNet lite".
Is this possible to do with an Arduino and some additional, of course cheap hardware? Can anybody support me with building a prototype of this for testing?
Thank you in advance and best regards,
Hans Guenter
Hello Hans and welcome to the forum.
I have been using BirdnetPi for a month or so. That uses Birdnet Lite configured to run on a Raspberry Pi. I have been getting good results.
I have the microphone or digital recorder plugged directly into the Pi. It doesn't make a continuous recording, it just records what it detects to be bird calls.
I have used the digital recorder to record bird calls in the field to SD card and then replay them into the Pi. I don't have experience of using Arduino for recording audio to SD card.
Cheers.
That is quite a big project. There are commercial products designed for that sort of scale.
Most Arduinos are too slow and have too little memory to be used for high quality audio capture. The Teensy 3.2 and 4.1 are sometimes used, but you might as well get a Raspberry Pi or the like.
So no... The Arduino is not the best choice for this. 
The "regular Arduino" is a little slow for real-time audio processing, the built-in analog-to-digital converter is only 10-bits, and no operating system so the is no "real" file system. (There is a library with limited capability that allows reading/writing to an SD card.)
You'd also need a "microphone board" or a mic preamp. And since the ADC can't read the negative-half of the AC audio waveform, the signal has to be biased. (The bias circuit is simple or if you build your own preamp you can build that in.)
...You may not want to record to MP3. As you may know MP3 is lossy compression and you MAY want more-perfect recordings.
If you edit an MP3 with a "regular" audio editor it gets decompressed. Then if you re-save as MP3 you are going through another generation of lossy compression and there is some accumulated quality loss every time. (AAC is also lossy compression but it's immune to accumulated degeneration.) There are some special-purpose MP3 editors that can do some limited editing (cutting, splicing, volume changes) without decompressing.
MP3 isn't "terrible" at higher bitrates (higher quality settings with less compression) and it can often sound identical to the uncompressed original. But standard practice is to do all of your audio recording and production in a lossless format and then if you want MP3, compress once as the last step.
Also, MP3 compression takes some processing power. The Raspberry Pi might be powerful enough to do it in real-time while recording but I'm not sure.
Of course the whole point of compression is smaller files and if you are making long recordings that's important. And, WAV (uncompressed) has a file-size limit of 4GB whereas most other formats don't have any "artificial" size limits.
I run BirdnetPi on a Pi4. I see in the file folders that there are both wav and mp3 files of detected bird calls. They are not long in recording time, maybe 5 to 10 seconds.
The Pi4 does the recording and analysis. I can leave it running all day and there will be many calls detected, identified and recorded. There's a great deal of work behind this from the creators -- Cornell University in the US.
Many thanx for your comments so far. I know about some commercial products, but they are mostly hardly available. And, I like to have a cheaper solution! The sound quality with mp3 is not a big problem. I made very good experiences with "old unused" smarthones running an mp3 recorder app, but they have limited recording time due to their "old" battery pack. So, I like to get a "roundabout 10 $ solution" plus case for a long time (two weeks?) recording. Might the Raspberry Pi Pico a suitable alternitive?
I think you would get a definite answer from this forum:
https://github.com/mcguirepr89/BirdNET-Pi/discussions
This is the BirdnetPi discussion area on Github. Someone has probably tried to do what you want to do. If you find answers please come back here and let us all know.
Yet, Cornell Uni managed, and have been doing it for decades. Sometimes kids excel by ignoring the "grown ups."
I would try this:
It holds only 40 minutes of audio. But it might be possible to use with a microcontroller, an SD card unit, an external mic if you want and a battery pack.
So the thing can hold 40 minutes of audio, the SD card can hold hours. The trick would be for the microcontroller to handle the copying.
This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.