Kokoro is great. I built an mcp for it a while back that has gotten decent traction - https://github.com/mberg/kokoro-tts-mcp if anyone wants to go that route
I've been using VibeVoice's ASR (speech to text) model quite intensively for the past month and have found it to be a lot more reliable and out-of-the box functional then Whisper, parakeet and other models. The fact that is has diarization built into to the model is a huge win in my book. Without that you have to run a different model just for that which adds significantly to the overall processing time vs VibeVoice which gives you reliably great results. Big fan.
I created a virtual voxel 64x32 app that emulates the 64x32 screen of my Tidbyt. Voxels are 2D SVGs building off David Aerne's delightful Heerich.js (https://meodai.github.io/heerich)
I realize this probably doesn't need to exist in this world but I've gotten a lot of job from it. My main current use is to create virtual displays on extra monitors that I can power from a pixlet apps (pixlet apps generate a 64x32 webp image this emulates).
It's now possible to leverage channels to create a simple HTTP server to wrap it. I created a agent http as a quick proof of concept to demonstrate this.
This mimics the API of Agent API (https://github.com/coder/agentapi) which has relied on clever terminal scraping to provide an http interface for claude code.
The TLDR is this enables you to run Claude Code in a headless way via API calls (perfect for AI code orchestration) while using your Anthropic subscription. To achieve this before you would need to use Anthropic's Agent SDK and API tokens.
I just created agent-http that leverages the channels feature to enable you to wrap claude code with a http api. This provides an identical API to Agent API (https://github.com/coder/agentapi) that relies on terminal scraping to achieve this. Now you can interact with claude code in a headless manner using your subscription. Previously I think you had to do this via the Agents SDK which relies on api token use.