Skip to main content

Jeff MacKinnon reshared this.


I guess with the influx of new users to the Fediverse, we're doing introductions now, so here's mine:

I'm a # and # enthusiast and developer. I'm also an # who's trying to create safe spaces for those who are recovering from high-demand control groups.

#

Jonathan Lamothe reshared this.


So this whole # # experiment has been a smashing success. This will be the last post I boost from my legacy accounts, and I'll be checking them with less frequency. If you've not followed the new account, now would be a good time.

I want to extend a special thanks to opencube and fosstodon for giving me my first taste of the #.


Okay, I've made some progress getting my #PeerTube instance up and running, but I'm running into some kind of database migration issue and I'm done beating my head against a wall for today.
I give up. I tried to start over, and this time I can't get yarn to install the software because of a wrong version of node. I was able to get around that with nvm last time, but not this time for some reason.

I guess I'll have to wait for a version of PeerTube that supports the current version of node.

Content warning: PeerTube issues



Trying to set up a #PeerTube instance. First problem I had was that the version of #node in the #Debian repositories was too old. After figuring out how to install the current version, I'm now faced with the problem of it being too recent.

error eslint-plugin-jsdoc@38.0.6: The engine "node" is incompatible with this module. Expected version "^12 || ^14 || ^16 || ^17". Got "18.7.0"
It's gonna never cause a problem!
On top of it all, I'm using a reverse proxy over a wireguard VPS because I have an 8TB drive I want to take advantage of for the file storage.
So, it turns out that the nginx configuration that comes with PeerTube, despite using a reverse proxy to port 9000 on localhost, still requires read/write access to the data directory. That doesn't complicate things at all.
@silverwizard Considering it. The only catch is that sshfs locks up the system for a non-trivial amount of time when the connection is lost. Also, up 'till this point, I've had it set up so that my local machines can SSH into the VPS, but not the other way around.

The workaruond I was thinking of was setting up an nginx reverse proxy at both ends.

Fingerprint Software reshared this.


I've been trying to be mindful of the # of the content I create. To that end, when I release a # on #, I've been (for the most part) taking the time to manually create # files for them, but it's a time-intensive process and some of my longer videos I simply haven't had the time to caption properly.

I've just been pointed at vosk, which can be used to automatically generate caption files for me. They're not perfect, but they're a decent stopgap measure until I have the time to do it properly.

Shout out to @Matthew Skala for bringing it to my attention.

3 people reshared this

Well, it looks like my machine isn't powerful enough to run the full training model, and I'm not interested in farming this out to the cloud. Luckily, there are simpler ones that can be run on a Raspberry Pi if necessary, and I was able to find a model that's within my machine's capabilities.
@Jonathan Lamothe (he/him) I find this "all our nothing" approach to computing these days annoying.

Whatever was wrong with "slow"?
@Ian Molton In this particular case, the "slow" option involves days worth of tedious manual work on my part, turning a hobby into a soul-sucking full-time job.
@Jonathan Lamothe (he/him) Sorry, that reads wrong...

I meant that software seems to be written with no software fallbacks in some cases, not that you were being personally lazy! πŸ˜€
@mattskala there’s no precompiled model?
And I've finally got all of my videos captioned. I've got to say, vosk did a surprisingly decent job.

This website uses cookies. If you continue browsing this website, you agree to the usage of cookies.

⇧