Hello, Which websites do you know and use to convert text to speech? Free, paid.
Google translate can, sentence at a time.
I remember seeing a few. Some let you save to a sound file.
Narakeet does pretty well.
The most natural sounding one that is free that I’ve come across is the immersive reader in Microsoft Edge browser. At least for the ones I’ve used it for…German and Spanish. You can choose from a variety of accents too.
There is also a program that uses the same API and let’s you convert text to speech independently of a browser, it produces an MP3 file and works from the command line as well: GitHub - rany2/edge-tts: Use Microsoft Edge's online text-to-speech service from Python WITHOUT needing Microsoft Edge or Windows or an API key
For English, the best AI voices I have heard come from: https://beta.elevenlabs.io/ They have samples on their website, e.g. This Voice Doesn't Exist - Generative Voice AI or this (angry) rendition of The Great Gatsby AI Narrates The Great Gatsby - YouTube
Thanks. I may look into that. For now having it in edge has mostly worked out for me. Plus I’m having to switch to Edge anyway for work =). But this could be useful so I’ve saved the link.
I think someone, maybe you?, had also pointed to something on azure that also had the “edge” voices as well. However, I think it may have costed money. It wasn’t super clear to me at the time if it was or not.
Indeed, the Edge browser uses the Microsoft Azure voices. They are identical to the best of my knowledge, the difference is that the paid version allows customization (SSML, pitch, speed etc).
An interesting tidbit is that LingQ actually uses Azure for TTS in Chinese Traditional, problem is that LingQ decided to put spaces between the characters when making a request. Since there are no spaces in Chinese, Azure interprets these as pauses, resulting in a stuttery mess in sentence mode or when reading a complete lesson. For most of the other languages LingQ uses AWS Polly. The regular Polly is pretty bad in my opinion, Amazon wants you to pay for their far superior but pricey “neural” voices, LingQ however doesn’t.
The LingQ voices since 5.0 and maybe slightly before sound pretty good for me…German and Spanish. I think the only fault I have with them is there is no pause between sentences, but oddly I think there’s a slight pause between phrases separated by commas. So the speaking still feels very rushed and not natural. Not sure if they’re just passing in sentence by sentence and therefore the between sentence spacing disappears or whether it’s simply the service doing that.
It also oddly goes to “English” accent whenever it encounters a word it deems not in the language. Edge/Azure keeps the speakers accent which is nice. I had a bit of a funny incident the other day…for some reason it didn’t automatically switch to German when I gave it something German to read (usually it does) but was stuck on Colombian Spanish accent. That definitely sounded odd. But overall I really like it. Will look into the elevenlabs one you posted.