natolambert / blogcaster Goto Github PK

I install everything right. but when It comes to execution, it fails. I am exporting an article via Notion. I tried changing image file names and keeping them as it is. still same issue. Your workflow is beautiful and I have a use case for this for my magazine website. would want to use it. Please help.

See docs https://apidocs.imgur.com/

Many voices struggle on longer sequences, so I should send my paragraphs per sentence for highest quality, or 1-3 sentences.

It would be more engaging if we could have a talking avatar. There are some open-source models and non-free ones as well.

https://github.com/natolambert/interconnects-tools/blob/6c917e46912603a5493d631799336bbef61b3a88/scripts/tts.py#L85

Might be boilerplate example code, but searching for that hash didn't find any results. If it is, you should probably remove it before someone runs up a bill on your account (remember to clean it out of your Git history too!).

Enjoying the blog, BTW.

• Leading spaces in text causes weird stuff before speaking, remove them
• Things like $1mil are hard to filter but don't work, maybe an LLM can rephrase to written word?
• Fractions don't work.
• Parenthetical's often sound really bad. Remove or rewrite them.

(Feel free to add, but this is my log book)

Considering the following system prompt:

Please perform the following task: translate the input into written word so a text-to-speech model can read it (things like fractions don't work well).

Examples include 1/4 to one quarter or $1.5m to one point five million dollars. Most dollar signs should be converted. When given a sentence, just replace those.

Logic will be to look for the following symbols in text: $, /, x, . (without a space after it)
Note: generate with 0 temperature for these :)

For generating short clips / adding captions directly to video. Would be great for sharing / replacing things like Headliner.

Also, move the file uploading to tts.py for easier workflow.

Make it so podcast listeners can easily see content by clicking on numbered figure in description.

1. Figure out which CDN to use.
2. Figure out formatting (markdown?)