As part of our podcast website migration from WordPress to Jekyll, I’m coding in a bunch of functionality that I’ve wanted for a while. The first one: a cli utility to fetch movie metadata into (semi-)structured data.
"overview":"In the post-apocalyptic future, reigning tyrannical supercomputers teleport a cyborg assassin known as the \"Terminator\" back to 1984 to kill Sarah Connor, whose unborn son is destined to lead insurgents against 21st century mechanical hegemony. Meanwhile, the human-resistance movement dispatches a lone warrior to safeguard Sarah. Can he stop the virtually indestructible killing machine?",
Check the help output for full and up-to-date functionality details, including a flag for interactive mode for when you’re less than confident that you’ll get the right answer. 🤓
You may be a Linux podcasting person looking for some ideas after a couple of years of Linux podcast production. You may be a fan of Decipher SciFi who wants to see what it looks like on the back end. In either case, welcome!
This is not meant to be the guide to doing it right, but just a record of the hackey way that we do it for reference. It’s like looking at someone’s Vim config; you don’t just copy the thing, but take what is useful.
Our usual recording setup is the two of us in the same room with sporadic remote guest appearances via Skype. This is the stuff we use and how (links to products on Amazon may include an affiliate code 🙂 ).
My partner and I both use ATR-2100 microphones. We settled on these for a few reasons:
Each has both XLR and USB interfaces, so we were able to continue to use them when graduating from computer to dedicated recorder.
They’re not overly sensitive so they work well when recording two people in the same room.
But, there is another. Since our mics and the system output from the Skype machine both go directly into the recorder, we need to have another microphone for our guests to hear us. We keep a Blue Yeti (which we do not recommend for the actual podcast recording) sitting in the middle of the table for this purpose, but whatever you have will probably do.
We started out recording directly into the computer with Audacity which proved to ultimately be less than reliable. Maybe this could be a fine enough affordable option for one person, I suppose, but if you’re recording two people in the same room you have a lot of pain-in-the-ass fiddling with audio configurations to look forward to, iirc (to aggregate separate USB audio sources). After a while and a few disastrous software failures in this mode, we finally got a dedicated recorder. In fact, I would actually recommend a dedicated recorder no matter what platform you’re on.
Zoom H5 recorder
So, this thing. This thing is pretty great! It can record four channels – two single-channel XLR inputs on the bottom, and a combined stereo input from whatever “capsule” is plugged into the top. With the X/Y capsule (it ships with this one) on top we are able to plug in remote guests from a Skype session on the nearest computer and record all three tracks separately.
Our recording setup
Okay, finally on to some software stuff.
The H5 records in WAV, and can basically only record input from the capsule (Skype) in stereo. So I want to get these tracks into mono FLAC files because it is lossless without being huge.
With the edit done, I can then run it through a final set of filters and cut in some intro and outro music (generously provided by Caelum Rale). Final export to 112kbps mono CBR LAME MP3 (plenty good enough for voice) with our id3 tags, add the thumbnail (Audacity doesn’t support this), and we’re done.
We use Levelator because it works, still, and is at least free as in beer. It’s super old and out of support but it works just fine for us, so here it stays. You can get Levelator here, but the link to the Linux version is broken. The Windows version works superbly in Wine though!
Levelator is pretty easy to use
The alternative these days seems to be Auphonic, a web-based service that gives you a few free sessions per month or somesuch and is reasonably priced for more. It seems popular and I hear good things.
Audacity doesn’t know how to write the cover art id3 tags, but it’s easily done on the cli anyway. eyeD3 is a good tool for this.
Our finalized MP3s get uploaded to Libsyn. Another good option I am happy to recommend is Blubrry. Either way, as an enterprising Linux nerd, you’d only be using it for file hosting because you’ll control the RSS feed itself from your own site? Right?
All kidding aside, people in the space continue to fight over whether you should control the feed from your own domain or relinquish control to the hosting service itself. You do you, but I prefer the former. And there’s a great WordPress plugin for that too mentioned below!
Our post for the episode then goes up on Decipher Scifi. The one good option here is WordPress(.org), really. How and where you do this is up to you but we run a multisite WordPress install on Linux on AWS and manage it ourselves. I do have a couple of tips re link shortening and Let’s Encrypt though if you’re into that sort of thing.
And then within WordPress, we use the only (and really good!) plugin for serving up podcast RSS feeds, Blubrry PowerPress. It’s never done us wrong.
Yeah don’t forget these! Both the audio and the website.
When the episode is up and out, we back up the raw tracks and the separate final edits of the tracks in FLAC, and the final cut in MP3. They go to Dropbox, Google Drive, S3, whatever, and a portable hard drive or two. The idea is to have both a local and remote copy if you value your backups.
The final edit files for one of our episodes
I wrote a small hacky BASH script that does my WordPress backups. It tars and gzips both the entire WordPress directory and the database dump and then moves it around with Rsync. One day I’ll do the work of finding/creating a better, more robust solution. Let me know if you can recommend one?
The free tier of this wonderful service allows for scheduling up to 30 posts, which is more than I ever do at one time. And now it can even take advantage of Instagram’s direct-posting feature for business accounts, if you have one.
Now since so many GIFs that I wish to post to Instagram are actually shorter than 3s, I automated everything above and here is the first cut of the script. To see if I made any changes since actually using it “in production” check the version I’m currently using in my dotfiles.
if[$# -eq 0 ]; then
Convertagif tomp4 with ffmpeg,looping it enough times toensure it meets
Instagram'sminimum video length limit.
infile|Avalid gif filetoconvert.IfgivenaURI,thisscript will
At work we noticed how the LastPass plugin on the new Firefox Quantum (due to limitations of the new plugin architecture?) no longer has copy a “Copy Password” button. LastPass support suggested that this might not be changing anytime soon. We rely on this a lot for different terminal-based work, so it was a sad revelation.
This bothered me enough that I made my own LastPass popup thing for my Linux desktop and as a bonus the workflow is also much faster than it was in the browser.
# Login to lastpass-cli one time and it will remember your email for the
lpass login email@example.com
And then run it there(no) or bind it to a hotkey(yes) in your window manager/whatever (I think this is under Keyboard settings in e.g. Gnome?).
When the menu pops up, just start typing the entry you want. It searches through both entry Name and Username. Whichever entry you select in the dmenu popup, the corresponding password will drop into your clipboard. Easy peasy.
By default as of this writing, lpass seems to just leave your password in the X primary clipboard forever (or until overwritten). It recognizes an environment variable however, LPASS_CLIPBOARD_COMMAND, where you can specify your clipboard command and arguments. This allows for a setting like the following
Which will allow one X selection request (i.e. a paste action) before this value is cleared from the clipboard. Hopefully the default will change in the future to be more secure? But there go in the meantime.
<edit> After further consideration it seems the environment variable trickery above will remain the only solution, so get used to it 😉