Hi, Is there a compatible microphone sensor module for raspberrypi that records in 16 bit?
Do you mean a module that records an audio file? Or otherwise, you could just get the pi to record in 16 bit with a standard mic attached: https://core-electronics.com.au/wired-miniature-electret-microphone.html
Also checkout this topic from a little while back: Use Raspberry pi to record audio
How do I attach the microphone directly to the board? Or do you mean through an analog to digital board? And if that the case what board do you recommend to achieve this without noise coming through the electret mic? Thanks
If you want recommendations it would probably help if we knew exactly what you want to do with this mike and it’s expected environment.
Electret mikes need power. If you chase through Core products to find some sort of data sheet you will find a little circuit showing how such power is connected.
Here is the datasheet for the similar mic with a breakout board: https://cdn-shop.adafruit.com/datasheets/CMA-4544PF-W.pdf
Sorry, this reply somehow got into the wrong thread. Please ignore
Indoor environment, voice recording with In a room 4m by 4m
I want to be able to capture everything said in a room, like a spy recorder.
Just want it clean from noise tho. Thanks
There’s a few option here for what we can use, I’d recommend going with the auto-gain microphone as it adapts the gain to match the ambient noise in the room, but it will require some soldering, external components and setup first. Otherwise, there’s a mini-usb microphone that’s not great in terms of being able to filter noise at a distance, but is an easy plug and play solution. If there’s anything else that we can do for you please let us know. Have a Merry Christmas!
Good luck with those requirements. Just ask anyone who uses a hearing aid.
The usb microphone has a lot of noise. Does the max9814 do 16bit?
“Random” noise would be pretty hard to get rid of. There are noise cancelling microphones but tend to be a bit large. They have access to the rear of the diaphragm so that “random” sound pressure impinges on both sides and thus is cancelled. Microphone then responds to whatever sounds are directed straight into it. In this case the babel of conversation in the room would become “random” and be cancelled. I don’'t think that is what you want.
The way out of this would be to use several units around the room and record individually or mix them first then record.
If these microphones are covert I think you may want to check the legality of this venture.
The MAX9814 is an analog device. The “bits” come later. (codec???)
Bob is definitely on the right track here, checking the legality of the project should be your first port of call.
A good method, although quite involved to remove noise would be to filter the signal before sending it through to the Pi, I would look into a High/Band pass filter to achieve this, there are a ton of contributing factors to implementing one of these to its full potential.
In regards to the 16-bit resolution of the mic, were you after a wide gain of frequencies on the input?
The mike is an analog device. It will not have a “bit” resolution as such. This, along with the sample rate would determine the ability of the digital system to accurately reproduce the analog signal after encoding and decoding.
The mike itself would have a sensitivity and frequency response quoted along with impedance.
If you are only after speech without noise a low pass filter would be called for. Something rolling off at 3 to 4 kHz. This may remove some of the timbre of the voices but you can’t have it both ways. To get rid of the higher frequency noise the higher transient and sibilant speech frequencies will suffer.
“Bits” (and pieces) seem to be the catchword these days in our so called digital world. Sound and indeed Radio are analog and there is no way around that. Even your so called “digital” Modem is analog once it heads off to the telephone line. Modem, short for modulator/demodulator.
Very good catch there.
I knew of the filters in theory but haven’t implemented them in any of my projects. Very interested to learn more!
Ah Filters. Almost a study in itself.
It would entail quite a bit of study to change that statement to “I know the theory of the filters” but well worth even a cursory look to get an understanding of what is going on. It would probably take months to delve too deeply.
There are hundreds of filter types in use. Broadly low pass, high pass and band pass consisting primarily of Inductance, capacitance and resistance in various configurations depending on lots of different criteria.
Probably the simplest you would have seen is an OP amp with a capacitor across the feedback resistor. This is a very simple low pass filter, The 3db roll off point is when the capacitor reactance equals the resistor value.
There are filters with very steep cut off curves. These usually require several stages or very high Q tuned circuits. The sharpest I have had dealings with at audio range frequencies was -3db points +/- 1.5Hz either side of centre and -60db 5Hz off centre. This was in a specialised bit of test gear. (built by AWA in fact).
Sometimes inductor values need to be quite large at lower frequencies and become impractical. There are circuits around called Gyrators in which a handful of OP amps can be made to look like a large inductor. In a nutshell if you have a network and you can make the current through it LAG the voltage across it by 90deg then you have an inductor. Conversely if you can make the current LEAD the voltage you will have a capacitor. This was used in the early valve days of FM generation where a valve would be across a tuned circuit modulated by audio to change the value of Cap/Ind.
Then you can go on with series and parallel resonance in RF circuits, they are all types of filter in their own way. The list in not quite endless but pretty long.
Getting to the digital age I think there are devices around where some reactance values can be altered digitally. May be loosely termed “digital” filters. I don’t believe how you do it makes any difference to the basics. I haven’t gone into it but it strikes me that maybe the easiest way to implement a low pass filter digitally would be to reduce the sampling rate at the encoding phase. Probably wrong but that is my idea anyway.