Albanian Videos

I want to make some videos using AI to help people learn Albanian. The videos will be entirely in Albanian but should be understandable. They should take an approach similar to Stephen Krashen's comprehensible input approach, where the scenes that are portrayed have enough context that you can infer what the meaning of each word is and how it's said. I want the videos to be describing a situation or an experience that you might be having and to convey a lot of information about diverse ways of saying things. I think you should not be limited to only understanding things at a particular level. You should be exposed to what people would say in the given context and then use that to try to understand as much as you can of what's going on. Through the visuals, understand what is going on more than trying to constrain what is being said to things that are simple or easy to understand.

For example, videos could be based on making something in the kitchen and going through the steps of making that. I think it's most compelling if it's interaction between more people, and they're asking things of each other, collaborating, and doing something out in the world together. You get to see what people ask each other, how they respond, and hopefully you can go through a lot of scenarios, but hopefully not boring ones that are often in language textbooks. You have to really decide on how to make interesting ones.

In addition to showing the scenes unfold and highlighting key objects or actions and trying to give the context on what is happening, I think it could make sense to also, before the different scenes, show certain vocab that will appear in the scenes and have visual examples of each of those words. That way, you can be primed with those and understand them as they come up.

Can use gpt-5.5 to write scripts and image prompts. Can use gpt-image-2 to generate scene images, and modify them to

Bringing Classic Novel Adaptations to Life

• Focus on high quality, ok to be somewhat manual
• Have ChatGPT Pro re-write novel as a graded reader
• Extract set of main characters
• Create a character sheet for each one with
  • Rendering of their likeness & wardrobe
  • Description of voice + choosing from one of the gemini voices that is suitable
• For each chapter
  • generate imagery to represent the scene(s) / key moments
  • Break out narrator from character dialogs
  • Generate audio for narration of the chapter
  • Extract timestamps of narration so we can time moments in music
  • Generate music for the chapter / scene with musical dynamics tied to key moments in the scene
• Use gemini 3.1 tts multi-speaker for narration
  • Pick a narrator and then have other voices for the main characters
• Use gpt-image-2 for the scene imagery