This is very early days in my Nerves journey. Does anyone know what route to take for getting good quality audio recording on Nerves (probably a Raspberry Pi)? Toying with the idea of building a long running bird song recorder for ecological research and wondered if this was at the “sure just use this library” end of the spectrum or at the “read these 8 specs then implement them as a C NIF” end?
Also, I have assumed “high end” is more than a jack/USB mike is capable of but maybe I am overthinking it?
I would say that the problem greatly depends on the hardware you are using. There are many types of microphones and interfaces but you will likely want to use a HAT with an I2S interface or use a USB sound card.
For the quality of the recordings, I think the biggest factor will be the actual microphone, as the characteristics (such as directionality, frequency range, etc) can vary drastically. I suggest you go with whatever is easiest to acquire and get started with (especially some sort of kit from Adafruit or Sparkfun, as they have good docs)
From the software side, fortunately it all looks pretty similar. Audio is typically handled in the kernel, so you’ll have to customize the nerves system for the board you’re using to add the driver. You might also want to add the alsa package for interacting with the drivers such as recording audio files.
There’s lots of resources and help in the community if you get stuck on this part
What’s your power budget look like for this application? A lot of fancy audio capture stuff is going to assume you’re either plugged in (studio etc) or only want to use it for a couple hours (live performance and similar)
The simplest (but also probably biggest + most power-hungry) approach would be to use a USB audio capture device or a microphone with one integrated (usually called a “USB mic”).
Other options:
use any analog microphone with an I2S ADC, for instance the Teensy Audio Adapter. You’d need a driver for it to work with Nerves, though.
use I2S microphones directly, for instance using modules like this breakout board. Two of these can be used on a single I2S port for stereo operation
The other neat thing that can be done with the tiny I2S microphones is using a bunch of them and then applying post-processing (similar to techniques used in radar) to form directional “beams” that can be adjusted after recording. This would naturally involve a LOT more data (dozens of audio streams) but could produce results that are impossible with a mono or stereo signal.
I have built membrane with the portaudio library for Nerves but never got to testing it. Should work fine for getting raw audio into Elixir for processing and media pipeline fun.
The challenge is almost certainly in getting a good microphone setup as mentioned above.
Thanks for this, with current progress and solar and batteries I am pretty confident I can meet the power needs within my budget (unless I am orders of magnitude wrong).
I am starting from a VERY LOW level of knowledge here so I will be starting as simple as possible. OTOH - sounds very cool!
