Podcast as a YouTube Video

Product

Critique 1 (a YouTube video with two art critics discussing the Gallery Catalog)

Process

Anthropic’s Claude is another of the LLM-based AI engines. Development activities moved to this new environment to see the possibility of creating a new podcast in a way that would have more flexibility.

There were several features that were desired.

Expanding to more than two voices in the conversation.
Choosing voices to better fit the conversation.
Editing to add, remove or modify the conversation.

The Claude Dialog

The conversation with Claude demonstrates the process of creating a podcast from scratch. Some of the dialog has been removed to shorten this discussion.

Inset portions are Claude’s responses. Bold text are the instructions provided to Claude.

The attached PDF document is a gallery catalog. It describes an upcoming exhibition at the NYC gallery. Can you give a basic description of the gallery show, based on the contents of this catalog?

This is the Fall 2024 exhibition catalog for the Knight-Brown Gallery, located at 248 Bleecker Street in New York. The gallery is presenting their 42-year tradition of showcasing exceptional mid-career artists, featuring five artists in this show:

Tad Nakamura (b. 1987) - A Japanese artist who blends traditional aesthetics like Ikebana with modern elements, creating striking still life photographs featuring flowers and vessels. His work shows the influence of both his Tokyo upbringing and European Old Masters techniques. Harry Jones (b. 1994) - A Boston native who creates abstract paintings using stripes as a visual language to document emotional timelines. His works, influenced by abstract expressionism, chronicle his life experiences through patterns of colored lines. Lulu Chen (b. 1985) - A Chinese-American artist from the San Francisco Bay Area who combines traditional Chinese painting techniques with Western styles. Her misty landscape paintings blend sumi-e (ink wash) techniques with modern abstraction, influenced by both her Chinese heritage and California surroundings. Fred Livingstone (b. 1977) - A farmer-artist from the Great Plains who creates vibrant paintings of agricultural subjects, particularly grains and crops. His work documents rural American life and landscapes while managing his family farm. Ken Nishida (b. 1952) - A Honolulu-born artist who practices Gyotaku (traditional Japanese fish printing) with a contemporary twist, combining this ancient technique with abstract expressionism to create unique prints of marine life. The gallery is managed by Hortense Knight-Brown and is open Tuesdays and Fridays from 11 AM to 4:30 PM (closed in August).

The cover image features Tad Nakamura’s “Tranquility 12” (2022), showing a white carnation in a gray vase against a dark background.

That’s very good. Can you create a deeper discussion, now with two people? I’d like them to have a lively dialog where they comment on both the works of art and the artists.

Here’s a dialog between two art enthusiasts discussing the Knight-Brown Gallery show:

Sarah: Did you notice how beautifully curated this show is? The way they’ve brought together five artists who are all sort of… straddling different worlds?

Marcus: Yes! Like Tad Nakamura’s still lifes - they’re technically photographs, but they feel more like Dutch Golden Age paintings. That piece “Tranquility 12” on the cover… the way he captures that single carnation against the dark background. It’s minimalist but so dramatic.

Sarah: And speaking of straddling worlds, what about Lulu Chen’s landscapes? They remind me of traditional Chinese scroll paintings but with this very California sensibility. That “Tomales Bay” piece really captures that Bay Area fog, but the technique is pure sumi-e.

Marcus: laughs It’s funny - someone could look at her work and Fred Livingstone’s pieces and think they couldn’t be more different, but they’re both essentially painting their homelands. His “Corn Harvest” is just… spectacular. The way he captures that golden light on the kernels, those dramatic vertical streaks…

Sarah: And you can tell he really knows his subject matter. That’s not just a painting of corn - that’s a painting by someone who understands corn, if that makes sense? The same way Ken Nishida’s fish prints show his deep connection to marine life.

Marcus: Oh, those fish prints! I love how he’s taking this super traditional Japanese technique - Gyotaku was originally just a way for fishermen to record their catches - and turning it into these sophisticated compositions. That piece “No schooling around!” with all those blue fish… it’s both playful and elegant.

Sarah: But can we talk about Harry Jones for a minute? His work is really the outlier here, but in a good way. Those stripe paintings are fascinating - especially that COVID piece. The way he’s turned nine months of pandemic life into those fractured, colorful columns…

Marcus: Right? It’s like a visual diary, but one that anyone can relate to. You can feel the isolation, the disruption. Though I have to say, I prefer his “Friends help each other” piece - those flowing, organic curves in earth tones. It feels… hopeful.

Sarah: You know what strikes me about all five artists? They’re all deeply rooted in tradition - whether it’s Jones with his Yale fine arts background, or Nishida with his Gyotaku, or Livingstone with his farming heritage - but they’re all pushing their mediums in new directions.

Marcus: That’s probably why Knight-Brown chose them. The gallery has always been good at finding that sweet spot between innovation and tradition. And notice how all these artists came to their current work through such interesting paths - Nishida was an educator, Livingstone still farms, Nakamura started as a graffiti artist…

Sarah: It really makes you think about the relationship between life experience and artistic expression. Like how Chen’s work reflects both her Chinese heritage and her California upbringing. None of these artists are doing just one thing - they’re all synthesizing different influences.

Marcus: And that’s what makes contemporary art exciting, isn’t it? When artists can draw from multiple traditions and experiences to create something new. This show really captures that energy.

Sarah: Absolutely. And the timing is perfect - showing this in Fall 2024, when we’re all thinking about both tradition and change. These works feel very much of the moment while still being connected to something deeper and more timeless.

Marcus: Well, shall we make another circuit? I want to take a closer look at those Nakamura pieces again. The way he handles light in “Spring Lust” is extraordinary…

Sarah: Lead the way! And maybe we can stop by the desk and see if they have any more information about Harry Jones’s process. I’m really curious about how he translates his diary entries into those stripe compositions…

Bravo! Beyond my expectations!

Thank you! I tried to capture the way art enthusiasts really talk - passionate but informed, making connections between different artists and works, and letting the conversation flow naturally from technical observations to broader cultural insights.

What I particularly enjoyed crafting was how Sarah and Marcus build on each other’s observations. They’re not just trading monologues; they’re having a genuine dialogue where each comment leads to a deeper understanding. For instance, when Sarah mentions Chen’s landscapes, Marcus naturally connects that to Livingstone’s work through their shared connection to place and homeland, even though their styles are completely different.

I also wanted to include some of the natural rhythms of real conversation - like Marcus’s suggestion to take another look at the Nakamura pieces, or Sarah’s curiosity about Jones’s process. These little details help make the dialogue feel more authentic rather than just an academic discussion.

Would you be interested in exploring any particular aspect of the exhibition in more depth? We could focus on a specific artist’s technique, or perhaps examine more closely how the curator has organized these diverse artists into a coherent show.

Using a New Audio Engine

ElevenLabs provides robust text-to-speech support, among other capabilities involving text and audio processing. This AI engine promises to add the flexibility that was limiting the NotebookLM production.

Formatting for ElevenLabs

ElevenLabs has a text-to-speech engine that’s quite flexible. The text for multiple speakers needs to be formatted specially.

Here is the dialog that generated the conversation in the proper format.

I’m going to use the current discussion to create an audio file. To do this, I need the dialog in a CSV format, with two columns (speaker, line). The line data should be in double quotation marks (i.e., “). Can you help with that?

Art Gallery Dialog CSV [file link] Click to open code

The CSV format preserves all the natural pauses and emphasis through ellipses and proper punctuation, which should be helpful for voice actors. Each line is properly enclosed in double quotes, and internal quotes (like artwork titles) use single quotes to avoid CSV formatting issues. Would you like me to add any additional formatting or metadata for the audio production?

Perfect. You’ve done it again!

Thank you! This was a great exercise in transforming natural dialogue into a structured format while preserving all the nuances that make the conversation feel authentic. The CSV format will make it easy to import into audio production software, and the careful attention to punctuation (especially the ellipses and question marks) should help voice actors capture the right intonation and pacing.

If you’re moving forward with the audio production, you might want to consider a few additional elements:

The contrast between Sarah and Marcus’s speaking styles - she tends to be more analytical and makes broader connections, while he often focuses on specific visual details The natural build of their enthusiasm throughout the conversation The subtle ways they respond to each other through verbal cues Would you like any other support for the audio production? I could help with suggestions for background ambient sound effects (gallery atmosphere) or notes about emotional tone for specific lines.

Editing the Dialog

The CSV-format file is easily modified in a text editor. The specific changes involved the following:

Adding a third speaker.
Adding text to introduce to critics and describe the situation.

The dialog with Claude finished with the following exchange.

I’m finished as the audio file was just what I needed. The CSV file slipped right into ElevenLabs software to generate the audio.

That’s fantastic to hear that the CSV format worked seamlessly with ElevenLabs! It’s a great example of how AI tools can work together - from generating structured dialogue to creating the actual voices. ElevenLabs is particularly good at handling natural conversational flows and nuanced emotional expression, so I imagine it captured those thoughtful pauses and enthusiasm in the art discussion quite well.

Was this for a gallery audio guide or perhaps a promotional piece? I’m curious about how the final audio turned out, particularly how ElevenLabs handled the more enthusiastic moments (like Marcus’s excitement about the fish prints) and the contemplative ones (like Sarah’s observations about tradition and innovation).

It’s satisfying when formats align perfectly like this - from natural dialogue to structured CSV to AI voice generation. Would you be working on similar projects in the future where I could help with dialogue or formatting?

Processing with ElevenLab

There are now ways, inside ElevenLabs app, to create a two-voice podcast from a script. However, in this project, we need three voices. This calls for a slightly-complex process.

Create a formatted file. This should be in a CSV format with a column labeled “speaker” and another “Line”. Here is an example:

speaker,line
James,“I’m James Cogsworth. Welcome to this week’s edition of Art Gallery Review. I’m here on Bleecker Street with two of my good friends. Sarah Jameson…”
Sarah,“Hi James.” James,“and Marcus Reed…”
Marcus,“and a good evening to you, too, James.”

Open the ElevenLabs app. Follow these steps to process your script.

Choose (from the menu) the “Audio Tools” and select “Voiceover Studio”.
Click on the “Create a Voiceover” button.
Supply a Project Name. Do not load a file! Click on the “Create voiceover” button.
Click on the gear icon in the upper-right corner. Click on “Import Script”.
Drag and drop, or browse, to load your script.
Change voices by clicking on the gear icon next to each name on the left of the audio tracks.

The subsequent processing should be familiar to anyone who’s used ElevenLabs applications.

Creating a YouTube Video

It’s OK to listen to the podcast. But it’s hard to recall the images in the catalog as they are discussed by the critics. The solution is to create a YouTube video with the podcast as the audio track. Still images serve as the visual elements to remind you of the catalog content.

Creating Additional Images

Several additional images were required to fill out the video.

Three participants: This image was created in Flux-1.
The Knight-Brown Gallery: This is an image of the gallery seen from across the street, at night, with the lights on in the gallery. The Google Gemini engine was tried first, but there were continuing problems getting the name of the gallery correct. Finally, the Flux-1 engine was used to create the basic gallery image. This image was then processed (using AI tools) in Photoshop to remove the existing artwork on the wall, place scaled Gallery Catalog images on the walls and add text to the gallery windows. The color of the tree leaves (on the left) was changed to better reflect the season (Fall).

Assembling the Components

Adobe Premier Pro was used to combine the still images and podcast audio. The result is a 4:28 long video ready to be loaded into and viewed on YouTube.