‘play HT’

Play.ht is an AI-powered voice generator and text-to-speech (TTS) platform that provides services for converting text into natural-sounding speech audio.

Similar to Eleven Labs.

The platform was initially launched as a Chrome extension in 2016 for listening to Medium articles and has since evolved into a more comprehensive tool for creating audio content. Play.ht allows users to emphasize specific words and sentences, alter rhythm, and create dynamic speeches, making it suitable for various applications such as podcasts, advertising, promotional materials, and audiobooks.

It supports over 100 languages and also has a Voice Generation API.

ttsmaker

TTS is text-to-speech, which is what Play HT is. There are several tools for this but the search volume for them is just growing and growing.

I haven’t seen much content out there comparing them or testing them on a number of tasks.

It’d be interesting to have a TTS Arena (like the LLM Arena) which compares them across the same benchmarks to see which platform is the best.

I think 2024 is the year a lot of us will chat with AI (like voice mode on ChatGPT, which is set to my action button shortcut) and listen to AI’s responses.

AI wearables are set to start getting into the hands of consumers next year which will be interesting to see how they integrate (or don’t) into every day AI-powered life.

‘AI wearable’ volume - growing but still small

‘codepal’

CodePal is an advanced AI-based code generation tool designed to assist developers in creating source code from a higher-level representation or plain language.

AI code generators are gaining significant traction already and I expect more non-developers looking at ways to dip their toes into building with code (generated by AI).

I really want to see educational content that focuses on non-technical people (like me) and helps them understand how to work with AI-generated code. Rather than teaching us to learn to code. I think there’s a new path emerging soon - learning to build with AI code.

or something catchier sounding 😀

‘learn to code’

People want to build things - or have them built without learning the pipes to build them. Maybe ‘learning to code’ is less enticing than ‘learning to build’ but often can be the same end path.

I started a no-code education company, Makerpad, which showed users how to build with no-code tools, instead of having to code.

The trouble with these things is that you still have to want to learn the system. You still need to do the work.

AI is making it much quicker but there are still so many hurdles working with code.

Using ChatGPT (or alternatives) to generate code, get it in Replit and live seems like the quickest route but again, has it’s hurdles.

‘replit’

Replit have a 100 days of Python course, which many people seem to drop off very early (I’m still on day 6). Which goes back to people just want to build the thing, not to learn the complete how before building.

If there were micro-courses of 5-10 days that were completely project-based, I think I’d be able to get further along.

Who knows, maybe I’ll create an email course that does something like this. Let’s see what time I have next year!

‘AI game generator’

AI-generated games are going to increasingly become popular. Whole games that choose your own adventure could be really exciting when AI is generating storylines, characters and obstacles on the fly.

We’ll see AI-first games start to become much more popular in 2024.

I’m not much of a gamer myself but this trend is hard to ignore.

Chat with your PDF is increasingly popular with many solutions out there at the moment. My favourite is PDF.ai. This behaviour will become more commonplace.

I wonder what the implications of this will be.

Information retrieval from large PDFs will get better too. Will consuming books change? Will we rely on this kind of interface for more types of data like spreadsheets?

That’s it for today!

Happy new year. We’ve got lots more to come.

‍