AI-Powered Weather Reports-Massive Update!

[RadioDJ Dude]

Wow, it SUCKED! I shredded my original Automated Weather Workflow and asked the bots to redesign it.

V2 is a huge upgrade!

Now, using all free AI resources (if you choose):
1. Weather forecasts are pulled
2. AI writes script based on custom prompts...infusing a friendly, conversational tone vs my dull, static output in v1
3. Eleven Labs records VO
4. VO is sent to your radio station's computer
5. My automated apps move, rename and process audio files to make them broadcast-ready
6. RadioDJ plays the report...hands-free, no file sync, nada nonsense.

BUT, a great addition: YOU! If you don't like the script or the VO, just tweak the script and send a email. The workflow automatically runs again. Powerful.

Don't think you're stuck with basic reports. Boring be gone! Wait until you hear these FUN weather options:

🔥 Meet Stormy, the snarky weather girl...perfect for morning shows.
🎤 A rhyming weather report—because why not?
🎬 An epic, action movie trailer-style forecast.


Put your eyeballs/ear holes on it here. Making weather fun again!

[JackRabbit]

Hey Jeff. On my system, I do not use the automatic silence detection setting, yet it seems like Variable-Duration files always use it regardless of the setting. My .wav file with the AI generated weather is padded with 3-4 seconds of silence before & after the speech.

What happens is RadioDJ totally ignores the silence on both sides. It starts the speech exactly where the "Next Start" cue point is set in the musicbed. That's OK, but my problem is that the next track starts playing immediately after the weathergirl says the last word. It sounds like that train wreck you always talk about. I would like for the musicbed to continue playing for a few seconds before stopping or fading-out. Any ideas?

[RadioDJ Dude]

You're exactly right! VDF files do apply the auto silence functionality whether it's activated or not in your AutoDJ settings. Thanks for the info!

One clunky workaround is to create a 1 second silent file and place it after the weather VO. Anymore than 1 second, in my opinion, sounds too loose. Set this file to Overlay File. I just tested it and it sounds great.

[JackRabbit]

I experimented using some "silence" files of various lengths, set as overlay. It didn't work for me, for whatever reason.

Then it dawned on me that I should just give it what it was looking for... noise. I made a wav file in Audacity that is literally just one beat from the "Generate / Risset Drum" menu. I "UN-amplified" it down to where I can barely hear it.

After the weather report file is generated, sox.exe will append the noise file to the end. It works great. RadioDJ lets the musicbed play on for a few seconds after the speech, and the noise is invisible to my ears.

I've been using just the free "google/gemma-2-9b-it:free" model with OpenRouter, and it's been generating some pretty good forecasts. OpenRouter is very generous with the free API requests, but there is a daily limit. I've only hit it once while testing different models. Somewhere over 40-50/day. Way more than enough for me.

Thanks for everything!!   

[RadioDJ Dude]

Wow, quite the workaround! BUT, I'm happy it's working for you!

Odd re. the silent file trick. It works for me. Did you import it without the silence settings? My file doesn't have any cue points.

Yeah, Open Router is awesome! They're quite generous.