Select, speak to edit, and see it on screen! A fcitx5 add-on makes Voxtype work like Typeless!

[rijuyuezhu]

Have you ever heard of Typeless? It is a voice input method with AI correction. You can also speak to edit existing text (for example, “rewrite it as an email”). Typeless does not support Linux, so I am trying to make voxtype work like it.

output.mp4

I have developed an add-on for the fcitx5 input method engine and made some modifications to voxtype. It now supports:

1. Voice input with excellent non-Latin support: press the hotkey (F9), then speak. The result will automatically appear after the cursor through the fcitx5 engine. This works even on GNOME! Since the process does not depend on keyboard emulation, it generates the result as-is, including non-Latin characters such as CJK or emoji.
2. Edit Mode: select text (just select it; you do not need to copy it to the clipboard), press the edit key (F10), then speak to edit it.

I have also implemented many interesting features in my voxtype fork, such as cancelable post-processing. I guess you might be interested.

If you are interested in the above workflow, take a look at my blog! To install it directly, see the installation guide.

• My voxtype fork: https://github.com/rijuyuezhu/voxtype.
• My fcitx add-on: https://github.com/rijuyuezhu/fcitx5-voxtype-bridge
• My blog: https://blog.rijuyuezhu.top/posts/efe0c0d6
• My development process (in Chinese, on an AI forum): https://forum.nju-aia.lsamc.website/t/topic/136