my-neuro
morettt/my-neuroReviewed by AI
Checked by RepoPicks Radar on
What is my-neuro?
my-neuro puts an animated Live2D character on your desktop that you talk to by voice or keyboard. Everything can run on your own computer: speech recognition, the language model, the voice (trained with GPT-SoVITS), long-term memory, image recognition of your screen, and extras like singing, live streaming on Bilibili, playing games with you and opening programs when you ask. You need Windows with an NVIDIA graphics card, Python and Node, and you follow the steps on the author's site. The main README is in Chinese.
my-neuro is free and open source (MIT) and can replace Character.AI, Replika, and ElevenLabs.
How easy is it to install my-neuro?
D · terminal, several tools or a specific graphics card · Windows
- Windows D
- Linux E
Windows is the best-supported system and the one the packaged installer targets, but it still needs Python, Node and an NVIDIA card, and `requirements.txt` says the CUDA version of torch must be installed by hand ("已注释,请手动安装CUDA版本"), plus a row of .bat files to start speech, voice and memory services. Linux is harder: `docs/deploy/linux.md` is for running from source only.
Coming soon: install in 1 click with the RepoPicks app.
my-neuro on GitHub
- Stars
- 1,443on Nov 10, 2026
- Downloads
- 27,84482 this week
- Growth
- +15stars in 6 days
Stars over time
Facts
- Runs on
- Windows, Linux
- You need to install
- Python, Node.js, NVIDIA graphics card
- To build from source
- Python, Node.js, NVIDIA graphics card
- Disk space
- 85 MB–254 MB · from the installer
- Accounts or keys
- OpenAI-compatible language model API (only if you choose a closed online model instead of a local one)
- Cost
- Free · Free and MIT licensed; the author takes donations on Afdian. A paid online model API is optional, since the project also runs local models.
- License
- MIT (Permissive)
- Can replace
- Character.AI, Replika, ElevenLabs
- Last change on GitHub
- Nov 7, 2026
my-neuro risk signals (5)
- Attentionsetup instructions offsite
The README sends you off the repository for the whole install: "本项目部署流程请看官网:[点我进官网](http://mynewbot.com/tutorials)", so what the installer downloads is decided outside GitHub.
AI review
- Attentiondownloads code at runtime
Top-level `installer.py`, `update.py` and `plugins-dlc/`, with `requests` and `pyinstaller` in requirements.txt, show the app fetches and installs extra parts after you start it.
AI review
- Attentioncontrols your computer
The feature list has "桌面控制:支持语音控制打开软件等操作" (voice control to open programs) and requirements.txt installs `pyautogui` plus a "屏幕点击工具" (screen clicking tool), so the AI can see the screen and click and type for you.
AI review
- Infocopies another creator character
The README describes the bundled character as "肥牛(fake neuro)它是一个抄袭neuro sama的角色" — a character copied from the streamer Neuro-sama — and the repository is named after it.
AI review
- Infovoice cloning included
The feature list includes training the voice model ("声音模型(TTS)训练支持,默认使用gpt-sovits开源项目"), so a real person's voice can be copied from recordings.
AI review
Evidence is shown in its original language.
What Radar could not check about my-neuro
- Whether replies really arrive in under one second, and on which graphics cards
- How much video memory and disk space the local speech, voice and language models need once downloaded
- What the installer on mynewbot.com and the plugins-dlc updates actually download and run
- Whether the Android app, the web panel and the game features work as the list claims
- How much a closed online model would cost if you use one instead of a local model