The issue with free tools is most of the ones I've tested make your outputs public.  They only allow you to make the outputs private when you're on a paid plan, and even then it's usually one of the higher tier plans.  I recommend testing a range of tools that do the job for free, then once you land on the right tool for the job you'll want to decide whether you still want free, or if you can stretch to paid, and it will probably come down to whether you're comfortable with the files you generate being completely public or not.  Note: "Public" on these sites usually means your generated output will be in their social feed and automatically put in view of other users.  It's not like other users have to go deliberately searching for your stuff, it'll just be in a public feed. 

Thanks for the comprehensive reply.

Looks like I will need to buy one.

Perhaps it would help if i explain the usage I want to turn 100 poems from text to speech in a mp3/4 file. I would like to be able to choose a voice that has some character rather than the usual robot.

Thats it

Look into NotebookLM

It's funny how we say "that's it" so cavalierly considering this would have been witchcraft to achieve only a couple of years ago 😂

gehenna:

 

Look into NotebookLM

 

It's funny how we say "that's it" so cavalierly considering this would have been witchcraft to achieve only a couple of years ago 😂

 

So very True, I still think its beyond amazing that I can push a screen button and have a real time face to face with my sister in England

On my iphone either in notes, onenote or word, I just tap the microphone button at the bottom of the screen.

On windows 11 in word, I click the dictate button. Otherwise, Windows+H brings up the speech menu.

For text to speech, on ios I use ‘spoken content’ (accessibility).

for windows (immersive reader) - it is built in, so select text and right mouse click and choose read aloud.  

https://huggingface.co/HKUSTAudio/Llasa-3B?

I had a use-case similar to this, where I needed to:
- Take a meeting transcript
- Summarize it (I can do this locally)
- Then get TTS of the summary for sharing. 

After trialing local TTS models, I ended up caving using eleven-labs - I know it's not free, but I just wrote some Python to take my output from Whisper, and chuck it up to them and get my audio files back. Worked pretty well. 

I need to look at local TTS models again soon, but need more compute in the homelab if I want any more LLMs.