Sowing Confusion Episode one - Bringing life to Spruce Openai

Planning for this podcast happened way before we even recorded the first episode. The show concept sat in my head since fall of last year and started becoming a reality when Annette indicated her interest in starting a podcast.

The six main episodes you will be hearing were taped between March and September, but pre-production and planning started in December 2024.

While it has been around for a long time, in the last couple of years, generative Artificial Intelligence (AI) has spread like wildfire to the everyday end user. Our third episode on November 26th will do a deep dive on the subject. Staff have serious concerns about AI but do acknowledge its presence and acknowledge that we need to have a working understanding of it if we want to be successful in our everyday career.

Lud does feel it is a drain on creativity and it is going to cause a lot of cognitive social-behavioral issues that we cannot even imagine at the present time. This will play out when the younger generation will age and Lud believes there will be a correlation to the prevalence of generative AI and an onslaught in cognitive and mental impairments. This does not include the issues we are seeing in social engagement at the moment with so-called "AI companions" that are being debated in real time even as we speak.

That said, while we would love to have real people in all of our podcast roles, we do not have the resources at the present time to do that. At the same time, Lud will always write and research the subjects himself without using AI because of it being prone to hallucinate. I will not use it as a writing assistant, because there are dead giveaways when you use it as one (em dashes anyone?).

That said, given tight funds, our announcer is generative AI. This though does allow for subtle commentary on the subject as the series moves on throughout our run. The question becomes how do you utilize chat GPT as your AI announcer when you are using the free version of Chat GPT, and only have ten minutes of access to the various voice characters each day (I have noticed recently in the last month, they may have increased that time limit)? The answer my friends is a lot of discipline.

Coaching Sessions with Spruce

For this segment I am going to share how the commercial cutaways were created. On every show when we leave and return from commercial Spruce says:

• "I don't know what the heck we're listening to, but we will be right back after I regain my bearings; after these messages."

• "We now resume the state of confusion, and I am in a state of perplexity or is that the city of perplexi, thank you very much. And here are your hosts, Ludnet and Anud, I mean Lud and Annette."

I am going to focus on the second bullet for this one.

Now how do you put something like this together, when you have to get chat GPT to say it in a certain tone, in a certain way, and you only have 10 minutes of the daily limit before it cuts you off and then you have to wait until the next day.

Rule #1 - Have the Script Entirely Ready and Up Before Logging In to Chat GPT

Know what you want written and how you want it written. Additionally, spell out how you want the AI assistant to do the voiceover. When I first did early recordings for the purposes of these posts, I captured them in a free open- source program called OBS Studio and I had on one screen my script on one-half of the laptop screen and had Spruce on the other side of the screen. That way you could see chat GPT in the video with me in the corner of the screen.

With the script in place for me to provide directions and chat GPT just logged in seconds prior to me giving Spruce the specific prompts, his allows for me to have the most bang for the buck during our session since our time is limited and will allow for multiple retakes in a very short time window. This specific recording was done in February when Spruce's program Chat GPT was more than willing to do emotional vocal ranges. Lud speculates over time that the vocal variety went away because Open AI saw a lot of usage in the unpaid membership using synthetic voices for podcast segments without paying for the premium features.

Keep the Recorder Running - Listen to the AI

If the AI did not read it the way you wanted, rephrase or indicate how you would like the script reread and be as specific as possible (e.g., a smidge slower, more sad, etc. and keep the recording going. You can always edit afterwards.

When finished hit the X on Chat GPT

Close chat GPT immediately once done. That way if you have a few minutes left and you need to do an extra line later, you are not using up that free time. If the time was used up, well I would just wait to the next day to continue.

Music

The musical elements put in were the music library from Corel VideoStudio proX8 software Lud owns. He would mix the video session into that software, cut out the audio, put it over music and then extract it as an audio file and would have the final product for his coming back from break segments.

Further show Stuff

As the show went on, Lud instead of using OBS Studio, went to Audacity and instead of capturing video, just captured Spruce's audio into that software. He then would still go into Corel and pull up that software to get his music. For ads like Launch Pad job club, he used the other voices in Chat GPT and had to do the production over several days due to the various characters and length of script. That is how the production elements of pre-recorded sound drops like audio bumpers were done.

Here is a video showcasing this item in action!