We’ve now got Tom reading our forecasts. This is an absolute clusterfuck of a setup, it’s so damn scuffed.
So Voxin doesn’t have a text2wav type utility, so what I have to have this do is create a virtual output device, record that for two minutes in another terminal, read the forecast, after two minutes stop recording, have FFmpeg trim out the silence, then append my intro. It’s scuffed as all hell but it works. Here’s a demo.
Well, that’s a good idea.
Yeah, and if the power goes out, the forecast wouldn’t change, and my Windows machines don’t auto-reboot when power is applied. I don’t want to be across the country when the forecast stopps working and not be able to fix it for weeks, so I run it on the cloud machine that runs the AzuraCast instance.
Oh no.
Well the problem here is that the script needs to run on a Linux machine, so I can’t use SAPI voices, otherwise I would. Problem is, my internet and power go out enough here that I can’t depend on a local Windows machine running a copy of Balabolka from command line and WinSCP or something enough to potentially use it for this.
Or maybe squeeze there arms? And also if Tom doesn’t work you could try a sapi voice? Like maybe David or something Or, you could use microsoft Sam?
Well Tom’s already having issues, if he doesn’t work consistantly then I’m gonna switch back to Festival, start breaking heavy objects over random people’s heads, and then probably just go sit silently in a corner somewhere.
aah damnit, I was loving the festivle tts, not that tom is bad or anything
That sounds a lot clearer now than when it did, and sorry for how messed up it is, I meen sorry it was hard to set up.
Leave a Reply to zlunglrg Cancel reply