Dec 22 2024 Update
Killed the project- ElevenLabs released the same feature and it's awesome.
Dec 3 2024 Update
Rewrote the scraper code in firebase functions only to have the deploys not support browers binaries :/
Will most likely move functions to Vercel Edge.
Dec 1 2024 Update
One app that I’ve always been a fan of was Speechify. Cliff’s (CEO) message is that it enables people with dyslexic read faster. I personally use it 1)on my walks 2) while making egg fried rice - more passive information osmosis while multitasking.
However, I’m reluctant to pay $29 USD a month for what feels like feature creep (e.g., 200+ voices, AI summaries, voice cloning). For a user like me, the north star metric is likely the amount of information absorbed with minimal interaction with a GUI. The 80/20 solution (closer to 80/1) I programmed on Saturday morning allows users to:
- Paste in a link
- Listen to or download the audio file at multiple speeds
And it costs a few cents.
Additionally, the text-to-speech models I’ve used sound much better than the in-house model built by Speechify (and support all major languages). Example from the latest Paul Graham Article on writing.
The site will be up soon.
Some features I’ve thought about adding were:
- Text-to-audio highlight sync
- Allow users to click on specific sections of text and have only that section read out
- Voice commands (for playback control, navigation, etc)
- Content management
- Audio streaming (near instant text to speech)
- Automatically filters out non-essential text (headers, TOCs, etc) from web content