Precorder

Psuedo-magically record audio from the past.

When improvising, you can't know when you're about to come up with something good. And you can't always reproduce what you came up with afterwards. And you shouldn't have to try to. With Precorder, you don't.

---

I like to improvise on the piano, but when I go to record what I just played, I might recall the "gist" of it, but I can't remember a lot of the variations that I came up with originally while experimenting. At the very least it can have a different "feel" the second time. And sometimes I can't remember much of it at all.

If I always started a recording before I played, I would end up saving a ton of long, largely mediocre recordings. I would think, "well, there's some good stuff in there." I wouldn't listening back to the recordings and edit each one down to contain only whatever good new stuff was in there.

"If only I could go back in time to record."

With this tool, you still have to hit a button to save a recording, so you don't end up saving lots of stuff you don't really want or care about, but you don't have to hit a button to record, so you can capture creative whimsy, post-whimsy.

#postwhimsy

Requirements

• A more-or-less dedicated Raspberry Pi or comparable device
• A more-or-less dedicated USB microphone (or if you want to use a non-USB microphone on the Pi, you'll need to buy a USB sound card (or perhaps a sound card "HAT"? (Hardware Attached on Top)), and you may need to jump through some serious hoops)
• Node.js and SoX (installation instructions below)
• Willingness to use the command line
• Who knows, because I haven't gotten this working yet (at least not on the Pi)

Setup

I'm not sure this thing works yet, so you probably shouldn't bother with it, but these are the approximate steps it would take:

• ssh into (or open a terminal on) the device you wish to use
• Install a recent version of Node.js. The best way is to use nvm; that way you can get the latest versions, and can switch versions easily later. You can run [their installer][nvm installtion] and then run nvm install --lts to get a recent, supported version.
• Install SoX with sudo apt-get install sox
• Clone the project with git clone https://github.com/1j01/precorder.git
• Enter the project folder, i.e. cd precorder
• Run npm i to install dependencies
• Run npm link to make the precord CLI available
• Plug in your USB microphone or sound card
• Find an ALSA device ID such as hw:0,0 by running arecord -l and looking at the numbers where it says card N and device N; the ID should be of the form hw:card,device where card and device are the numbers from that command (maybe it could need plughw in place of hw, idk)
• Set some environment variables, using the device ID for PRECORDER_AUDIODEV: e.g. export PRECORDER_AUDIODEV="hw:0,0" ; export AUDIODRIVER="alsa"
• Run npm start
• Wait for it to say "wrote chunk file"; it might show an error at this point
• If it worked, run precord 1min to save an audio file in data/output/ (and make sure it's an actual recording you can play)

TODO

• Get it working!
• Set up npm start as a daemon process and allow running at startup; maybe use npm stop to stop it
• Make the CLI actually respect the time given
• Make the CLI accept an output file path
• Handle stdout specially: maybe add an alias (-); don't output other stuff along with the audio file data
• Polish up and publish to npm