Sound Interfaces and Microphones, a tricky combo to get?

[instanttim]

I started with my RPi4 and because of shipping prices I begrudgingly decided against getting a EM272-based mic from micboosters. I got a Rode Lav GO, and I found a usb audio interface that I thought was reasonable from Plugable. After setting this up, I recorded some test sounds inside my home and the RMS was low, the peaks were low, everything was low, it sucked.

Doing some research, I was thinking maybe the Plugable adapter simply wasn't supplying sufficient bias voltage for the mic. So I got a Sound Blaster 3. Similar results. So then I tested things a bit more scientifically, using that microphone connected to a Sony digital recorder I have, which yielded results that were ~12x better. Then I used the SB3 with the same mic on a Mac, thinking maybe the Pi was the problem. Poor results.

So poking around here I saw some people post links to micverve which is selling EM272 mics at prices that were much better than I could find (why didn't this show up in google and shopping searches?!) But I'm not at all confident the better mic will fix my problem since it seems like my issues stem from the USB audio interfaces -- even though neither if them is cheap junk afaict.

Links:
https://rode.com/en-us/products/lavalier-go
https://plugable.com/products/usb-audio/
https://us.creative.com/p/sound-blaster/sound-blaster-play-3

[stevedee]

Not sure if you had a specific question, so what follows may be of zero value to you, or anyone else!

BirdNet is a sampling system which attempts to identify bird calls within a very limited range. Its been trained using thousands of field recordings (not high quality studio recordings). Its not very interested in low frequencies or those much above 12kHz, and can tolerate a certain amount of 'noise'.

Some users appear to spend an enormous amount of effort & cash trying to obtain the highest 'quality' (whatever that means).
The truth is that the main factors limiting a BirdNet system are: #1 environmental noise (e.g. wind, rain, traffic, barking dogs & so on), and #2 system sensitivity.

Almost any microphone will give acceptable results, so long as you can't hear any obvious audio distortion. Sensitivity can be the tricky one, so this may require some testing. If you have to buy mics & modules, even cheap ones, this can work out expensive.

The very cheapest option is probably to buy one of those stubby USB mics that is about the same size as a stumpy memory stick.

For those that understand soldering and Ohms Law, a cheap solution is to buy electret mic capsules, screened cable, jack plugs & low cost mic USB modules, and put it all together. Check the mic drive voltage (higher is normally better).

I've been running BirdNet-Pi for a few years, and through various stages of testing, I currently use a low cost electret mic capsule mounted into the end of a hollow bamboo cane, connected to a cheap 'non-branded-white-box' USB audio module. This system is mounted in my loft with the mic cane pointing out through the roof, and it typically detects 28-32 species each day (...although the neighbours dog does good impersonations of ravens, partridges & whooper swans).

The bottom line is: don't throw lots of money at the problem by buying increasingly expensive mics & modules. Analyse the problem, and be sure to adjust capture gain settings for the correct audio device.

[instanttim]

You know, I was so focused on details I didn’t realize I didn’t really ask a question. It’s really I was looking for any insight on parts that worked well together for folks here.

Your post has super useful information, I guess I should have also added that my project is a bit more involved because I’m streaming the audio so I can play it on speakers inside my house — bringing the outdoors inside. So I am interested in making sure I keep the noise level down even though the bird identification part might still function.

With that said, my setup was not picking up enough audio to do anything with. I don’t have the test numbers in front of me, but normal sounds (conversational voice for instance) were barely showing up a foot away from the mic. The same mic on a dedicated recorder worked fine.

That would seem to indicate the mic is the problem. But I replaced that and same problem. So now I ordered a different mic. Some combination will hopefully work!

thanks for your informative post!

[stevedee]

I have no idea how much you know about this subject, so forgive me if I'm saying things you already know.

The Rode + Plugable combination should work fine: both are 3 pin TRS, which means the wiring is compatible.

Low sound levels on the RPi suggest that the 'alsamixer' settings need some adjustment;
With your BirdNet-Pi system running, go to Tools > Web Terminal
Enter your username & password (as requested)
Type: alsamixer
Hit the F4 key
Your display should look like this;

Set capture level initially to 100%
Hit the esc key
Select menu Spectrogram
What can you hear now?

If its still the same as before, go back & re-open alsamixer, then hit the F6 key
Select: USB PnP sound device, then the F4 key and set capture level.

Test again; what do you hear now?

[instanttim]

Thanks, I appreciate the help. I did already crank up the settings in alsamixer to 100. I don't tactually have BirdNet installed yet as I ran into these questions when setting up the hardware and testing my devices. I'm installing the full project now to see if I can test with real world data.