Suggestions for USB microphone systems

[DD4WH]

Hi,
it seems it is difficult to find good and reliable mics for using with BirdNET-system.

I would like to share a working setup with high performance that is used by many users:

UPDATED on April, 24th, 2022

• Mini Pluggy EM272 Omni mic - https://micbooster.com/clippy-and-pluggy-microphones/65-clippy-em172-microphone.html
• If you dare to solder a mic, you can build a very high performance mic for about 10,- EUROS following this instruction: Suggestions for USB microphone systems #39 (comment)
• for wind protection, search for "Lavalier windshield" or "Lavalier foam windshield" on your favorite shopping platform
• Do NOT use a cheap non-brand USB audio sound card, they often produce unwanted whines and/or hissing noise. Use a good well-known brand USB audio sound card. These can be purchased for 10-20$.
• Be careful that the 3.5mm plug type of your USB audio sound card matches the one at your microphone [TRS or TRRS]: https://en.wikipedia.org/wiki/Phone_connector_(audio)
• Almost every USB audio sound card available is of the MONO input type. So you can only connect ONE microphone to the BirdNET-Pi with those sound cards, THIS soundcard is a little bit more expensive, but is of the Stereo input type: https://www.ugreen.com/products/usb-audio-adapter-with-3-5mm-2rca
• ALWAYS use a USB extension cable! Do not plug in the USB sound card directly into the RPi: this will introduce electromagnetic interference noise from the RPi into the microphone, because the mic is too near to the interfering chips on the RPi
• ready to go! Patrick @mcguirepr89 has programmed the system in such a smart way that the USB sound card is automatically detected!

In summary, this setup will give you good performance for a total cost of about 90$ (including the USB extension cable and postage)

Best regards, Frank DD4WH

[CaiusX]

Thanks @DD4WH - I will follow with interest!

Also interested in your DIY approach discussed in #11, please keep me in the loop.

Best

[andrew0123ncd]

The following are six hints on what to search for in a USB receiver.
You Want a Quality Mic. I frequently express "shop at Guitar Center, not at Office Warehouse".
1.Pick a Huge Stomach Condenser Mic.
2.Search for a 20-20,000 hz Recurrence Reach.
3.Zero Inactivity Earphone Observing.
4.Think about a 24 Bit Mic.
5.Look at the Graph.
6.Have Studio Will Travel.
Also here https://lumbuy.com you can buy what you want.

[gerritzen]

4.Think about a 24 Bit Mic.

How do you enable that in BirdNET? My 24 Bit Sound card runs with 16 bit and I couldn't find a way to change it.

[saidone75]

Given that an excellent mic has 80 dB SNR switching to 24bit makes very little sense indeed.

[Vollpflock]

Wouldn't it make sense to use 24 bit if you need to amplify it in post?

[croisez]

I just refresh my memory by reading following article: https://www.wikiwand.com/en/Audio_bit_depth .
I think that the most interesting in using 24bit i.o. 16bit is the dynamics of the signal. Your signal can be safely amplified, that without the fear to be clipped if the signal source become too powerful.
But, in our case, I don't think that we have a very changing dynamics audio signal. This could be the case when recording a symphonic orchestra, for example. But in our case ... no.
For me, the fundamental is to capture the audio source with the highest quality and fidelity, reducing the noise at its maximum.
I would personaly prefer a recording chain using a very good mic giving very high SNR along with a 16 bit adc, instead of a poor mic connected on a professional 32bit-float adc

[ghost]

I asked some experts in bird recording, they recommend this one --> https://de.rode.com/microphones/videomicntg . But it is really expensive, costs 240€. It has a high pass filter which can be usefull for bird recordings and also a special kidney form characteristic.It can be attached to any tripod. I plan to install the RPI also on the Tripod including some self made accumulator and place it in the forest for some hours.

€.

[ghost]

Yes thats clear, but better have a radius of 25m in one direction as a round radius of 5m, thats what I understand. Sorry when i ask, are there OTHER automatic ID systems ? (cannot find any...)

[DD4WH]

Yes, there are many others:
search for
bird identification machine learning
and you will find hundreds of entries . . . its a hot topic :-)

[DD4WH]

interested to hear about your experiences with the directional mic! Let me know when your system is ready and going!

[ghost]

Yes, but see you the Price? I have to wait and save money...:-)

Ok I know what you mean.

[ghost]

When i have the microphone I plan to record a short video about the System.

[ghost]

Here is my current Microphone, i can really NOT recomment it, it hast a lot of noise resp. bad SNR.It has cost 3€ in China.

[jcaesar]

The annoying part about this thing is: It works fine when I plug it into my notebook. More noise than voice when talking normally to it plugged to a raspi.

[ehpersonal38]

The annoying part about this thing is: It works fine when I plug it into my notebook. More noise than voice when talking normally to it plugged to a raspi.

Have you tried with a USB extension cable? Like the original post mentions there can be a lot of noise if plugged directly into the pi.

[jcaesar]

Yes. 3m extension cable. It changes the noise (dampens the lower frequencies a bit, adds a whine at ~900Hz), but overall doesn't really help.

[ghost]

Other suggestions for Microphones?

[DD4WH]

Are you an advertising bot? If not, please explain which mic you think is suited best for our application and explain, why so.

[DD4WH]

There seems to be a misconception about the difference between omnidirectional microphones (most Lavalier mics, electret capsules like the one mentioned above in this thread etc.) and directional mics (cardioid, supercardioid, hypercardioid, shotgun mics). Let me explain why I think there is no advantage of using directional mics for the specific purpose of detecting birds with BirdNET-system based on a Raspberry-Pi:

Imagine two mics placed in the field, one DIRECTIONAL MIC (maybe a shotgun mic) and one OMNIDIRECTIONAL MIC (like the lavalier mic mentioned above):

• in the frontal direction, the omnidirectional mic and the directional mic have the same probability of detecting a bird (limited by the selfnoise of the mic, not so much the sensitivity
• Note that the directional mic has NO advantage in hearing birds in frontal direction, because it has no additional noisefree gain in that direction
• the directional mic has a disadvantage in all other directions, because it is directional, thus having ATTENUATION in other directions than the frontal
• the omnidirectional mic has equal sensitivity in all directions
• So, IF the radius is 25m in the frontal direction for the directional mic, it is also 25m for the omnidirectional mic (given equal selfnoise and sensitivity), but it is largely reduced for the directional mic in all other directions
• Now we can also look at signal-noise-ratio (SNR)
• The SNR will be better in the directional mic, because the "noise" (this "noise" will often be other birds you might be interested in!) from the sides and behind is being rejected by a certain degree

So, in summary, a directional mic is NOT more sensitive to the frontal direction than an omnidirectional mic. [please note that it is different with parabolas!]

• BTW: we have not talked yet about wind noise in shotgun mics! It is much higher than in other mics, so you would need to cover such a mic with a blimp and/or a dead cat which makes the system much bulkier and more expensive

If you ask an ornithologist (I am an ornithologist, BTW), she/he will recommend a directional mic for field work in situations where you have control over the direction of your mic and can actively point the mic to the bird AND your main interest is a good and clean recording of ONE bird species. BUT: this is not the case in BirdNET-system:

• in BirdNET-system, I am interested in a system listening to ALL the birds in ALL directions and have them identified
• And this calls for an omnidirectional mic
• BTW: BirdNET is capable of identifying simultaniously singing birds in a morning chorus for example

MY RECOMMENDATION FOR a mic in a BirdNET-system:

• use an omnidirectional mic with the lowest selfnoise that you can afford
• this can be achieved by using the EM272 electret capsule with a very low self noise of 14dBA (SNR 80dBA) --> see above for a link to a Lavalier mic based on that capsule --> you might notice that the Rode mic mentioned above has even a little bit HIGHER selfnoise rated as15dBA (79dBA SNR)
• my field experience with the EM272 based microphones is that they are excellent and they are reasonably cheap!
• if money is no matter to you, buy a superduper & ultralow selfnoise & field-tested & robust Sennheiser MKH8020 and a USB interface with 48V phantom feed (for example Focusrite Scarlett 2i2 --> I have not tested this with the RPi so far!) to have the maximum performance :-)

LISTEN HERE FOR SOME of my RECORDINGS MADE WITH THE EM272 (each capsule costing about 15$):
https://github.com/DD4WH/SASS

But your mileage may vary and your goals might be totally different in using the BirdNET-system, so use whatever microphone and USB interface you want to.

But please report your real life results and experience here rather than speculations how a specific mic might perform with BirdNET ! :-)

[ghost]

Hello Frank, nice to have you here when you have experience in bird recording. I agree and complain with all what you write. So what is the best mike with the best range/price with USB connection? Here is what I understand so far :-)

The capsules you recommend are fine (i think I have even some of them at home), but the raspi has no analog input. That's the reason I thought about to spend the 200€ for the røde NTG. I first ordered one for 40€ and heared later on YT that it has a bg-noise-signal and canceled the order. The Sennheiser is toooooooo expensive :-) lol. I have another problem that I want to develop and battery pack later and every mA from the sound-card / microphone counts. The SNR of your self build microphone system is really good, like the whole microphone seems to ideal.

So by all what you say and recommend I am still by mine røde videomic NTG which i want to order for 3 reasons:

1.) It has an USB with low SNR
2.) It matches on my unused video/microphone stand which I have.
3.) For the dirction from the house wall and in the field/forest.

Perhaps it would be better to continue this discussion on the Partik's BirdNET-Lite implementation.

[DJM0]

Good advice, in my setup I have a busy road at the front of the house and behind where the birds are is a small section, using a cheap omni lavalier I have a lot more base noise than a cheap shotgun (NTG4+) which also has the benefit of eliminating all of the road noise.

[nilspupils]

@DD4WH: Thanks for the excellent writeup and suggestions. Much more clear now! Will definitely check out the EM272.

[DD4WH]

@Christoph-Lauer
Your drawing is misleading and not correct, because both mics have the SAME distance of hearing.

Violet is the omnidirectional EM272 microphone (or any other omni)
all the other lines is the Rode shotgun mic.
the grey lines are different attenuation levels in dB. [inner circles: high attenuation, outer circles: low attenuation]
As you can see, there is no difference in front of the microphone.

[ghost]

Could be that you are right, but could also be that the directional microphone has more sensitivity and amplification so it overreaches the omni in the one direction? Thought they have 2 or more microphones inside and do a kind of analog beamforming in the directionals. I am not a totally expert in microphones, i will ask the one who recommended the rode ntg.

[DD4WH]

well, the very last thing I will try to convince you of that is to point you to the datasheets of the two mics, which prove exactly that . . . have a look at sensitivity and selfnoise of the two mics . . . and NO, there is no "2 or more microphone beamforming magic" inside standard directional mics (that kind of beamforming is restricted to very specialized applications like eg. expensive conference mic systems etc.)

https://de.rode.com/download/videomicntg_datasheet.pdf
https://micbooster.com/datasheets/EM272Z1.pdf

[ghost]

What about the absolute values? I mean the SPL sensitivity, this where the first values i tried to compare before i asked some "experts". The dB polarplots are always nomalized to 0db. The basic question is clear --> what has the best radius where the birds can be heared.

UPDATE: Found this on the rode datasheet: -26 dB re 1V/Pa (50mV @ 94dB SPL) ± 1dB @ 1kHz

[DD4WH]

Exactly! And you will see in the EM272 datasheet that it is entirely comparable to the Rode in terms of selfnoise and sensitivity. So they will cover the same bird distance (in the front). To the sides the EM272 will cover a similar distance, in the Rode everything to the sides and behind is attenuated.
So you have to draw a revised version of your picture ;-)

[ghost]

Ok, i see it now... remains the question for the USB interface and microphone mount (I want to place the raspy incl. the batteries on the mic stand). I see on the schematics that the capsule needs 3V power supply. The 100 bugs for the microphone you propose with this capsule is still a lot of money, means that there are other USB Microphones to compare.

[ghost]

What about this Microphone: https://www.amazon.de/FIFINE-Kondensator-mikrofone-Studioqualität-Microphone-Schwarz/dp/B07QC5W7G9?th=1 o this one https://www.amazon.de/dp/B08CVP2HXP

[ghost]

Found a German page the Edutige EIM-001 and also the Rode videomic is recommended.

https://oagstrenzfeld.wordpress.com/beitraege/vogelstimmen-aufnehmen-fuer-einsteiger-april-2019/

https://www.edutige.de/wp-content/uploads/2021/01/EDUTIGE-Catalog-09-2020_ENG.pdf

[ghost]

Here are also recommendations (In German): https://www.slashcam.de/info/Mikrofon-fuer-Vogelstimmen-429198.html

[ghost]

Microphone: https://www.thomann.de/de/the_tbone_em_9600.htm
USB DAC: https://www.thomann.de/de/focusrite_scarlett_solo_3rd_gen.htm

Σ 150€

[ghost]

https://wiki.linuxaudio.org/wiki/hardware_support

[DD4WH]

BirdNET-system also works in STEREO out-of-the-box! Thanks Patrick, for this nice feature!
I tested BirdNET-system with a stereo low noise setup with two EM272 electret capsules in AB40 stereo setup and an ultralow-noise balanced mic line driver (by Elliot ESP: https://sound-au.com/project93.htm) fed into a standard cheap (about 25$) UGREEN USB audio interface: the recordings are of high quality: will test more with this setup and post a DIY suggestion in the next weeks HERE in this thread describing how to build this [at the moment the prototype is powered by a 9V battery, I have to build a power supply for this, so this will take some more time]

For the moment, the solution in the first post in this thread has proved very reliable and produced nice results! That is what I tested and can recommend. Probably it will work with any cheap USB audio interface available and the lavalier mics based on the EM272. I would not trust so much other lavalier mics available on the internet (if you did not thoroughly test them or you have a reliable data sheet with a figure of the self-noise at hand).

[ghost]

I ordered some cheap used devices listed above. Must have a directional microphone because I live on a street and must hide out the cars. I think the algorithm didn't note the stereo, I have some BG in Machine Learning, the core of the deep neuronal network is using spectrograms (mono) and the net topology for image recognition the spectrograms. Hopefully the USB device will work on the RPI.

[mcguirepr89]

A few versions back, I changed the default REC_CARD variable for installations to use "default," which in the desktop version of RaspiOS is the pulseaudio sound server. (pulseaudio is explicitly checked for during installation now, even though it is included in the base image I recommend for users, so that BirdNET-Lite-system works on the raspios-lite_arm64 images, too. That is all still in development).

What is standard, however, is that pulseaudio will choose what is appropriate (almost always) based on application needs. I set the default channels to 2, since the pulseaudio sound server can also abstract streams into virtual channels -- allowing support for channels that the hardware may not (i.e., the device might record in mono, but will be split into two anyways since that's what is set (it supports up to 32 channels in most cases, so if you do some serious recording with very specific channel allocations, these can be used, though it would require some know-how.)
PulseAudio is awesome

"By default, PulseAudio is configured to automatically detect all sound cards and manage them. It takes control of all detected ALSA devices and redirects all audio streams to itself, making the PulseAudio daemon the central configuration point. The daemon should work mostly out of the box, only requiring a few minor tweaks."

[blogdemoi]

"I set the default channels to 2"

So, if I am using a mono EM272 mic plugged in a Ugreen usb card, should I set this to 1 or 2?

[croisez]

I have just looked down under the hood, looking how audio files are used by the Birdnet classifier.

It appears that even if the audio file chunks have a stereo format, it will be opened as a mono audio file by librosa.load() function.
See /usr/local/bin/server.py at line 201: the parameter "mono=True" is given to the function.
This means that you may define CHANNELS to value 1 or 2, it will not influence how the analyzis program will do its job.

Practically, as your microphone is monophonic, imagine that your USB sound card has a stereo input, which I doubt, configuring CHANNELS=2 will make the recording of a stereo content, but with same information on both L and R channels.

My advice would be to keep CHANNELS=2.
This is working well on my setup

[blogdemoi]

"My advice would be to keep CHANNELS=2.
This is working well on my setup"

Thank you.

[robinsandfort]

Hi there,
we use the new audiomoth usb firmware: https://www.openacousticdevices.info/usb-microphone to get the
https://www.openacousticdevices.info/audiomoth running on the raspi.
I´m testing the μMoth: https://github.com/OpenAcousticDevices/Datasheets/blob/main/uMoth_Datasheet.pdf
They work great and can be set with all the nice filters you would wish for.

[dbrianhendrix]

I'm not sure I'm at the stage of having a permanent solution running all the time. I want to see how well it works with Audiomoth as a microphone as well as a TRRS microphone. I'll also be using one of several portable power banks I have. Also, my assumption here is that the Raspberry Pi 4B has to be setup with WiFi so you have access to it outside via the web portal http://birdnetpi.local/.

[dbrianhendrix]

Also, thanks for sharing! I didn't realize there was a Pi-Microphone HAT available -- just another "gadget" to play with my wife would say!

[dbrianhendrix]

Microphone ready . . . check!
Deadcat on microphone . . . check!
Microphone mounted on small tripod . . . check!
Microphone cable attached to USB Sound Card . . . check!
USB Sound Card attached to USB cable extender . . . check!
Cabling plugged into prepped Raspberry Pi 4 . . . Waiting for USPS deliver today!
Power "brick" ready to power Raspberry Pi 4 . . . Waiting

So my "rig" is ready. I just need to create software on microSD card and plug it into Pi and power up Pi and configure. However, today I see that there's a new BirdNET-Analyzer-Pi github site for the Pi version of new model. Not sure if I can use that yet instead of BirdNET-Pi.

Maybe @DD4WH can share!

[CaiusX]

Hi

Suggest you go with the PI repo at this stage - Analyzer is in its infancy, any experience/ideas/epiphanies from Pi will be included in Analyzer when appropriate!

[louispires]

Any further feedback on the best way to get sound into a Pi Zero 2?

[ghost]

[DD4WH]

@robinsandfort
I would not use the AudioMoth because the audio quality is low. The SNR is not competitive and there are better solutions for a similar or slightly higher price. I have compared the new version of the AudioMoth to the old version and to other recording devices and I cannot recommend it. Even attaching an external high quality microphone to the AudioMoths´ new version does not improve the SNR. Using the AudioMoth for ultrasound recording is even worse, so not recommended.

[nilspupils]

So after the BirdNET system seems to be very stable, i want to work on the sound quality for the recordings. So far i have been using a UGREEN External USB Sound Card Adapter (Type CM477) which works perfectly, and a AGPTEK Lavalier microphone type (AC02B) seems also all right. The first microphone (cheap chinese) had a very loud hum and on secondary inspection a faulty ground connection at the capsule.
Although i am pretty impressed by the detection quality with this setup, the signal level in the audio files seems to be quite low (almost looks like a straight line in the waveform view). First step will be replacing the microphone capsules with better ones and i thank you @DD4WH for the recommendation of the EM272, which seem to be excellent. For reasons of availability i chose another capsule, the
PUI Audio AOM-5024L-HD-R. http://www.puiaudio.com/pdf/AOM-5024L-HD-R.pdf
I will report again after the first tests in the following days.
Best, Nils

[nilspupils]

Low hum could be grounding issue?

[ghost]

That's the reason I opened it and all is connected correct.

[DD4WH]

@Christoph-Lauer
In order to use a stereo mic, you need a stereo input. Many inputs of laptops etc. only have MONO inputs (3.5mm socket has GND, INPUT, and OUTPUT, NOT 2 inputs . . .). Also many input 3.5mm sockets are of the TRSS type (4 connectiosn instead of three) and do not support plugging in a stereo microphone. Maybe this is the cause of the problem???

[ghost]

You could be right, the mono mic works.

[FirstTenor]

I have the original Sony version of this - with exactly the same problem. I believe that the issue here is mains hum being picked up on the lead between mic and Audio interface. I inserted a ‘ground loop filter’ (designed to block this - and it filters everything except the 50 Hz hum!)…. But 3.5mm is ‘unbalanced Audio’ and susceptible.

I just plugged in a ‘proper Audio setup’ - Focusrite Scarlett with a studio condenser mic and importantly - balanced XLR leads. …. No hum now and I can hear the birds. The trick now is to reduce the cost as I dont plan on leaving a $200 microphone outdoors

[ghost]

Found some tips for the microphonation:

https://www.audubon.org/news/a-beginners-guide-recording-bird-vocalizations
http://oldbird.org/mike_home.htm
http://soundbirding.org/index.php/which-recording-device/

[bytos]

Primo EM272M Special Version - improved RF Noise Protection
https://micbooster.com/microphone-capsules/278-1099-primo-em272-m.html#/36-single_2_4_matched_capsules-single/97-type-with_wires

A new capsule exclusive to Micbooster with vastly improved RF Noise suppression. Incorporates new circuitry to make the capsule much more resistant to RF noise from Wi-Fi devices, cell phones and other equipment which generates radio frequency interference.

[bytos]

@DD4WH I downloaded an audio file of yours, but can't remember the link.

It was good enough that amplification via software of the sound wave didn't show obvious signs of noise (low noise floor).

Could you link to one of your audio files recorded using the latest setup you recommend?

Clippy Mic (EM272M, TRS) => UGREEN Audio Interface (C-Media HS-100, TRS) => Cable Creation's USB Extension Cable => Raspberry Pi 4

[Strandmann]

Hello, I would like to share my experience because it was really difficult to find an inexpensive and good sound card with a stereo input.
Maybe it will be useful for someone who is looking and also wants to record sounds in stereo.
The AXG ADA-17 makes no noise and can be purchased for under €15 from Reichelt, for example. In conjunction with two AOM-5024L-HD-R and an old jack cable for video transmission, which I use as a connection between the microphone capsule and the sound card, I achieve a (in my opinion) clear audio signal and was able to capture over 2000 voices yesterday, March 5th, 2024.
The location is Bützow in Mecklenburg-Western Pomerania, Germany
I wish you successful crafting and lots of bird songs (:
https://www.reichelt.de/soundkarte-extern-stereo-hq-usb-2-0-axg-ada-17-p346294.html?&trstct=vrt_pdn&nbc=1
https://www.reichelt.de/elektret-kondensator-mikrofonkapsel-aom-5024l-hd-r-p285350.html?&trstct=pos_0&nbc=1

[arne1921KF]

Great project, even better community. Thanks to you all, I got my birdNET-Pi up and running. The only thing which really bugs me is the buzzing and humming I cannot get rid of when connected via any AC/DC adapter with the very old webconference microphone I had in a drawer. When I am using a power bank, there is no noise at all. I suspect the mic's cable shielding is terrible.

Hardware setup:
Raspi 4B 4 GB
USB extension cable
UGREEN USB soundcard (C-Media HS-100B chipset, EAN: 6957303837243 - currently on sale for below 6 €), TRS connector.
Sigma CM 20 microphone (with TRS connector)
Audio Card set to plughw:CARD=Device,DEV=0.

I also have a RODE VideoMic Me-C (which I use with the Merlin App or BirdNet App when in the field, usually), If I use the Me-C (Audio Card set to plughw:CARD=MeC,DEV=0, using a USB C OTG adapter to connect it to the same setup as above) there is basically no humming, so I am relatively sure it's the microphone - or rather, the cable - itself.

FTR, neither microphone works with Audio Card set to default, for some reason. If I connect them both, I can even switch between them by changing the audio card. But together or alone, pulseaudio does not seem to recognise each of them, which I find curious.

[stevedee]

Common mode noise is a frequent problem when running a Pi from a mains adapter and using an analogue microphone with either an analogue/USB adapter or an analogue card. You can sometimes use a digital voltmeter to measure the AC or DC voltage between your supply earth/ground and the 0Volt line/common on your Pi; you may see a voltage of several hundred milliVolts.

The fact that you have no problems with a PowerBank supports this idea.

I use an 0.47uF non-polarised capacitor between earth/ground and the Pi common to minimise mains hum.

I don't know where you are or your national mains supply configuration, but in the UK a capacitor or a direct short works well. (see my notes: http://captainbodgit.blogspot.com/2023/11/birdnet-pi-portable-operation-mics-bird.html)

[arne1921KF]

After a cursory reading of web sources, I am under the impression you connect GPIO Pin 6 with the capacitor in between to a ground, i.e. either the ground in the outlet socket or another (like, e.g., a radiator)?
(Just FTR, I'm currently in Germany. I'm thinking about plugging together several others, which I'll ask colleagues and friends to deploy them in Germany, Switzerland, Italy and Estonia. It's a pity that BirdNET-Pi isn't under active development at the moment - I definitely have to check for the forks!)

[stevedee]

If your mains plug/socket only has live & neutral connections, then you may find a 'radiator connection' will work.

Basically the power supply negative is modulating the Pi common/0Volt line (& therefore the microphone common) with noise, so if the metal pipework in your house can neutralise it, great!

Personally, if you are asking other people to use these systems, I would try using a low cost digital mic. This one works well for me: https://thepihut.com/products/mini-usb-microphone

[arne1921KF]

Excellent advice. Also, FYI: using GND did work, and most of the buzzing is gone.

[croisez]

Did you try this kind of audio cable noise filter? This stops EMI and RFI. https://www.amazon.com/VSKEY-Anti-interference-Telephones-Equipment-Appliances/dp/B075G2P5SW Le dim. 17 mars 2024 à 23:00, æ ***@***.***> a écrit :

…

Great project, even better community. Thanks to you all, I got my birdNET-Pi up and running. The only thing which really bugs me is the buzzing and humming I cannot get rid of when connected via any AC/DC adapter with the very old webconference microphone I had in a drawer. When I am using a power bank, there is no noise at all. I suspect the mic's cable shielding is terrible. Hardware setup: Raspi 4B 4 GB USB extension cable UGREEN USB soundcard (C-Media HS-100B chipset, EAN: 6957303837243 - currently on sale for below 6 €), TRS connector. Sigma CM 20 microphone (with TRS connector) Audio Card set to plughw:CARD=Device,DEV=0. I also have a RODE VideoMic Me-C (which I use with the Merlin App or BirdNet App when in the field, usually), If I use the Me-C (Audio Card set to plughw:CARD=MeC,DEV=0, using a USB C OTG adapter to connect it to the same setup as above) there is basically no humming, so I am relatively sure it's the microphone - or rather, the cable - itself. FTR, neither microphone works with Audio Card set to default, for some reason. If I connect them both, I can even switch between them by changing the audio card. But together or alone, pulseaudio does not seem to recognise each of them, which I find curious. — Reply to this email directly, view it on GitHub <#39 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AANRFSIMCVZZGXLNLS6ELWTYYYG6PAVCNFSM5GFGNFL2U5DIOJSWCZC7NNSXTOKENFZWG5LTONUW63SDN5WW2ZLOOQ5TQOBSGEZDMMA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[alexbelgium]

Hi, it didn't work for me : it added an additional interference band on my spectrogram. I guess however it all depends on the type of noise that must be removed

[SharonD214]

These are interesting just clip on the cable to the microphone?Sent from my iPhoneOn Mar 19, 2024, at 3:25 AM, croisez ***@***.***> wrote: Did you try this kind of audio cable noise filter? This stops EMI and RFI. https://www.amazon.com/VSKEY-Anti-interference-Telephones-Equipment-Appliances/dp/B075G2P5SW Le dim. 17 mars 2024 à 23:00, æ ***@***.***> a écrit :

Great project, even better community. Thanks to you all, I got my birdNET-Pi up and running. The only thing which really bugs me is the buzzing and humming I cannot get rid of when connected via any AC/DC adapter with the very old webconference microphone I had in a drawer. When I am using a power bank, there is no noise at all. I suspect the mic's cable shielding is terrible. Hardware setup: Raspi 4B 4 GB USB extension cable UGREEN USB soundcard (C-Media HS-100B chipset, EAN: 6957303837243 - currently on sale for below 6 €), TRS connector. Sigma CM 20 microphone (with TRS connector) Audio Card set to plughw:CARD=Device,DEV=0. I also have a RODE VideoMic Me-C (which I use with the Merlin App or BirdNet App when in the field, usually), If I use the Me-C (Audio Card set to plughw:CARD=MeC,DEV=0, using a USB C OTG adapter to connect it to the same setup as above) there is basically no humming, so I am relatively sure it's the microphone - or rather, the cable - itself. FTR, neither microphone works with Audio Card set to default, for some reason. If I connect them both, I can even switch between them by changing the audio card. But together or alone, pulseaudio does not seem to recognise each of them, which I find curious. — Reply to this email directly, view it on GitHub <#39 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AANRFSIMCVZZGXLNLS6ELWTYYYG6PAVCNFSM5GFGNFL2U5DIOJSWCZC7NNSXTOKENFZWG5LTONUW63SDN5WW2ZLOOQ5TQOBSGEZDMMA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: ***@***.***>

[croisez]

yes. and for a such low price, give it a try ! Le mar. 19 mars 2024, 13:33, SharonD214 ***@***.***> a écrit :

…

These are interesting just clip on the cable to the microphone?Sent from my iPhoneOn Mar 19, 2024, at 3:25 AM, croisez ***@***.***> wrote: Did you try this kind of audio cable noise filter? This stops EMI and RFI. https://www.amazon.com/VSKEY-Anti-interference-Telephones-Equipment-Appliances/dp/B075G2P5SW Le dim. 17 mars 2024 à 23:00, æ ***@***.***> a écrit : > Great project, even better community. Thanks to you all, I got my > birdNET-Pi up and running. The only thing which really bugs me is the > buzzing and humming I cannot get rid of when connected via any AC/DC > adapter with the very old webconference microphone I had in a drawer. When > I am using a power bank, there is no noise at all. I suspect the mic's > cable shielding is terrible. > > Hardware setup: > Raspi 4B 4 GB > USB extension cable > UGREEN USB soundcard (C-Media HS-100B chipset, EAN: 6957303837243 - > currently on sale for below 6 €), TRS connector. > Sigma CM 20 microphone (with TRS connector) > Audio Card set to plughw:CARD=Device,DEV=0. > > I also have a RODE VideoMic Me-C (which I use with the Merlin App or > BirdNet App when in the field, usually), If I use the Me-C (Audio Card set > to plughw:CARD=MeC,DEV=0, using a USB C OTG adapter to connect it to the > same setup as above) there is basically no humming, so I am relatively sure > it's the microphone - or rather, the cable - itself. > > FTR, neither microphone works with Audio Card set to default, for some > reason. If I connect them both, I can even switch between them by changing > the audio card. But together or alone, pulseaudio does not seem to > recognise each of them, which I find curious. > > — > Reply to this email directly, view it on GitHub > < #39 (comment)>, > or unsubscribe > < https://github.com/notifications/unsubscribe-auth/AANRFSIMCVZZGXLNLS6ELWTYYYG6PAVCNFSM5GFGNFL2U5DIOJSWCZC7NNSXTOKENFZWG5LTONUW63SDN5WW2ZLOOQ5TQOBSGEZDMMA> > . > You are receiving this because you were mentioned.Message ID: > ***@***.***> > —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: ***@***.***> — Reply to this email directly, view it on GitHub <#39 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AANRFSK62LDW22NFDQCYDN3YZAWCNAVCNFSM5GFGNFL2U5DIOJSWCZC7NNSXTOKENFZWG5LTONUW63SDN5WW2ZLOOQ5TQOBUGAYTMMQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[SharonD214]

Thanks I certainly will!Sent from my iPhoneOn Mar 19, 2024, at 8:55 AM, croisez ***@***.***> wrote: yes. and for a such low price, give it a try ! Le mar. 19 mars 2024, 13:33, SharonD214 ***@***.***> a écrit :

These are interesting just clip on the cable to the microphone?Sent from my iPhoneOn Mar 19, 2024, at 3:25 AM, croisez ***@***.***> wrote: Did you try this kind of audio cable noise filter? This stops EMI and RFI. https://www.amazon.com/VSKEY-Anti-interference-Telephones-Equipment-Appliances/dp/B075G2P5SW Le dim. 17 mars 2024 à 23:00, æ ***@***.***> a écrit : > Great project, even better community. Thanks to you all, I got my > birdNET-Pi up and running. The only thing which really bugs me is the > buzzing and humming I cannot get rid of when connected via any AC/DC > adapter with the very old webconference microphone I had in a drawer. When > I am using a power bank, there is no noise at all. I suspect the mic's > cable shielding is terrible. > > Hardware setup: > Raspi 4B 4 GB > USB extension cable > UGREEN USB soundcard (C-Media HS-100B chipset, EAN: 6957303837243 - > currently on sale for below 6 €), TRS connector. > Sigma CM 20 microphone (with TRS connector) > Audio Card set to plughw:CARD=Device,DEV=0. > > I also have a RODE VideoMic Me-C (which I use with the Merlin App or > BirdNet App when in the field, usually), If I use the Me-C (Audio Card set > to plughw:CARD=MeC,DEV=0, using a USB C OTG adapter to connect it to the > same setup as above) there is basically no humming, so I am relatively sure > it's the microphone - or rather, the cable - itself. > > FTR, neither microphone works with Audio Card set to default, for some > reason. If I connect them both, I can even switch between them by changing > the audio card. But together or alone, pulseaudio does not seem to > recognise each of them, which I find curious. > > — > Reply to this email directly, view it on GitHub > < #39 (comment)>, > or unsubscribe > < https://github.com/notifications/unsubscribe-auth/AANRFSIMCVZZGXLNLS6ELWTYYYG6PAVCNFSM5GFGNFL2U5DIOJSWCZC7NNSXTOKENFZWG5LTONUW63SDN5WW2ZLOOQ5TQOBSGEZDMMA> > . > You are receiving this because you were mentioned.Message ID: > ***@***.***> > —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: ***@***.***> — Reply to this email directly, view it on GitHub <#39 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AANRFSK62LDW22NFDQCYDN3YZAWCNAVCNFSM5GFGNFL2U5DIOJSWCZC7NNSXTOKENFZWG5LTONUW63SDN5WW2ZLOOQ5TQOBUGAYTMMQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: ***@***.***>

[BAM-Team]

Best one I tested (more than 10 different types) is still
Boya BY-LM40 USB-Lavalier-Mikrofon
excellent Mic :-D

[SharonD214]

Thanks this is far better that the one I had before!Sent from my iPhoneOn Mar 23, 2024, at 2:50 PM, Carsten ***@***.***> wrote: Best one I tested (more than 10 different types) is still Boya BY-LM40 USB-Lavalier-Mikrofon excellent Mic :-D —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: ***@***.***>

[wotipka]

Greetings to the brain trust. A year or so ago I saw a project (Hackaday or some such) about someone using a MEMS mic array to create a location grid of where the birds were on their property. I was wondering if that was a fork of this project or if anyone else had heard of it. Currently building a birdpi for my lovely wife who is all about the birds.

[DMontgomery40]

I don't think this is what your talking about, but it's a commercial product using BirdNet-Analyzer, very precise GPS, and MEMS mics, and if you had even a medium sized property, it would show you different activity as you moved it around. Again not what you're talking about, but, I have one (in addition to 3 birdnet-pi's cause i have a problem) and it definitely has a use case.
https://www.birdweather.com/
https://app.birdweather.com/ (this is the full map where they show it all together, if you so choose to volunteer your bird data)

[winedarksea]

Greetings to the brain trust. A year or so ago I saw a project (Hackaday or some such) about someone using a MEMS mic array to create a location grid of where the birds were on their property. I was wondering if that was a fork of this project or if anyone else had heard of it. Currently building a birdpi for my lovely wife who is all about the birds.

This is actually on my to-do list to make an open source project that does this sound triangulation and sound classification (like the infamous Shot Spotter but more general). I have even designed a circuit board for making a single point 4 mic array but it's probably going to be a year before I get anything out.

It's a well studied problem with books and academic papers and commerical products, but not a lot of open source work.

For now:
http://opensoundscape.org/en/latest/tutorials/acoustic_localization.html

And this might be the hackaday you mentioned:

https://hackaday.com/2023/12/30/localizing-fireworks-launches-with-a-raspberry-pi/

[ampsu]

Its a hard task. Ive played around with it but there are quite a lot variables in play. I have stations that record with timecode but its hard to calibrate everything. BirdNET detection isnt that reliable. Eg. I am monitoring one single species with a confidence threshold of 70%, one station detects the sound, other detects with 65% and data is instantly unreliable because it doesnt count the sound as a detection. When I lower threshold, I get a lot of junk data. Right now I am focusing on gathering a lot of sounds in order to build a new model for a certain species. Maybe it helps. Whole idea is automated system so I wouldnt have to label files manually.

[wotipka]

This is actually a pretty comprehensive sensor package for the price. I at this point, this would get her up and running while I play with my own setup and the MEMs mic setup as that technology progresses. Thanks for the heads up. kw

…

On Fri, May 3, 2024 at 8:47 AM Colin Catlin ***@***.***> wrote: Greetings to the brain trust. A year or so ago I saw a project (Hackaday or some such) about someone using a MEMS mic array to create a location grid of where the birds were on their property. I was wondering if that was a fork of this project or if anyone else had heard of it. Currently building a birdpi for my lovely wife who is all about the birds. This is actually on my to-do list to make an open source project that does this sound triangulation and sound classification (like the infamous Shot Spotter but more general). I have even designed a circuit board for making a single point 4 mic array but it's probably going to be a year before I get anything out. It's a well studied problem with books and academic papers and commerical products, but not a lot of open source work. For now: http://opensoundscape.org/en/latest/tutorials/acoustic_localization.html And this might be the hackaday you mentioned: https://hackaday.com/2023/12/30/localizing-fireworks-launches-with-a-raspberry-pi/ — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***>

[wotipka]

Well, wouldn't the data for the mems just be metadata? In other words, you get the recording from a single, high quality mic, but then extract level data from the mems? Would just need 4 values to consider with the audio file. Kw

…

On Fri, May 3, 2024, 10:06 ampsu ***@***.***> wrote: Its a hard task. Ive played around with it but there are quite a lot variables in play. I have stations that record with timecode but its hard to calibrate everything. BirdNET detection isnt that reliable. Eg. I am monitoring one single species with a confidence threshold of 70%, one station detects the sound, other detects with 65% and data is instantly unreliable because it doesnt count the sound as a detection. When I lower threshold, I get a lot of junk data. Right now I am focusing on gathering a lot of sounds in order to build a new model for a certain species. Maybe it helps. Whole idea is automated system so I wouldnt have to label files manually. — Reply to this email directly, view it on GitHub <#39 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABGTI37SVABHQT3T6P6JVVLZAOKXZAVCNFSM5GFGNFL2U5DIOJSWCZC7NNSXTOKENFZWG5LTONUW63SDN5WW2ZLOOQ5TSMZQGYYTQOI> . You are receiving this because you commented.Message ID: ***@***.***>

[BAM-Team]

Hi, it seems it is difficult to find good and reliable mics for using with BirdNET-system.

I would like to share a working setup with high performance that is used by many users:

UPDATED on April, 24th, 2022

• Mini Pluggy EM272 Omni mic - https://micbooster.com/clippy-and-pluggy-microphones/65-clippy-em172-microphone.html
• If you dare to solder a mic, you can build a very high performance mic for about 10,- EUROS following this instruction: Suggestions for USB microphone systems #39 (comment)
• for wind protection, search for "Lavalier windshield" or "Lavalier foam windshield" on your favorite shopping platform
• Do NOT use a cheap non-brand USB audio sound card, they often produce unwanted whines and/or hissing noise. Use a good well-known brand USB audio sound card. These can be purchased for 10-20$.
• Be careful that the 3.5mm plug type of your USB audio sound card matches the one at your microphone [TRS or TRRS]: https://en.wikipedia.org/wiki/Phone_connector_(audio)
• Almost every USB audio sound card available is of the MONO input type. So you can only connect ONE microphone to the BirdNET-Pi with those sound cards, THIS soundcard is a little bit more expensive, but is of the Stereo input type: https://www.ugreen.com/products/usb-audio-adapter-with-3-5mm-2rca
• ALWAYS use a USB extension cable! Do not plug in the USB sound card directly into the RPi: this will introduce electromagnetic interference noise from the RPi into the microphone, because the mic is too near to the interfering chips on the RPi
• ready to go! Patrick @mcguirepr89 has programmed the system in such a smart way that the USB sound card is automatically detected!

In summary, this setup will give you good performance for a total cost of about 90$ (including the USB extension cable and postage)

Best regards, Frank DD4WH

Try out the Boya BY-LM40 USB-Lavalier-Mikro
Until now I didn´t find any better mic. And I tested 10...20 differen types.
Cheers, Carsten

[alexbelgium]

Have you tried a mic based on EM272 ? It's currently with Boya By-LM40 the two bests I've tested.

The top 3 graphs are taken at a max 10min interval in my garden this evening (low noise) ; the bottom two other were previous mic : a very basic Dahua camera rtsp feed gave incredible results (it has noise cancelling embedded) and a smartlav+ mic was very expensive but not super nice

To be noted that it did not seem to change that much in terms of detection quality for common birds

Second test this morning, indeed Clippy EM272 and Boya LM-40 are very comparable :

This can even be seen from spectrograms which is very interesting as Boya LM-40 specs is up to 10khz and the others' up to 20khz but it still show signal until the 15khz used for the model (I don't have clippy on it, but smartlav was very near) :

[alexbelgium]

Edit, I've also tested the HyperX Quadcast on another thread based on a feedback, and have settled on a Clippy EM272 mic with a Rode AI aux/usb interface - looks to be the best possible system for a reasonnable price.

After a couple weeks of analysis it seems:

• Best entry system (< 50€) : Boya By-lm40 (30€) + deadcat (10 €)
• Best middle system (< 100 €) : HyperX Quadcast (80€) + deadcat (10€) ; but I'm concerned about the longevity of this mic outside
• Best high end system (< 150 €) : Clippy EM272 (55€) + Rode AI micro trrs to usb (70€) + Rycote deadcat (27€)

[pe1mew]

A suggestion for a USB microphone is this one: https://duurzamedealtjes.nl/products/staande-microfoon-usb
I have used it now for 1 week and have good results.