Manual mode can now work like a "real" voice changer: it converts your voice live as you speak, running about 1 to 2 seconds behind you.
For now, this works with the 5 female preset voices, and only in English, Russian, and Korean. The rest of the presets and languages are coming soon. Custom voices are not limited this way: custom voice training is available to everyone, in every language, today.
This runs on your own computer (graphics card or CPU), so it may slow down your game while it is on.
It uses only one fifth as much of your time, and with any plan (Shy and up) it is unlimited. This is separate from the text-to-speech engine, so its limits are separate too.
You can use it with a custom designed voice too, on the Casual plan or higher. Preparing a custom voice is a one-time setup of about 1 to 2 hours, depending on your graphics card. The app is busy while it runs, but the rest of your computer stays usable; games may just run slower until it is done.
Also fixed a bug where the local engine would not work at all. If you tried Local voice before and got an error, it works now.
Speak as long as you want in manual mode
Fixed a bug where long manual recordings still got cut off after 5 minutes. The app now quietly splits your speech at natural pauses and stitches the audio back together, so there is effectively no limit.
More natural pauses
The pauses in your speech now carry through to your converted voice, so it sounds less rushed and more like you.
Your balance shows up reliably
Fixed a case for Steam users where launching the app before your internet connection was ready could show the wrong balance for that session.