@xiangxiang103: A bro from Bilibili made a cyber girlfriend, so immersive! A fully local AI girlfriend, no internet, no API key. VAD: Silero VAD v5 STT: Whisper LLM: local llama.cpp TTS: Qwen3-TTS All four models packed into 1…
Summary
Introduces a fully local AI girlfriend project made by a Bilibili developer, integrating four models: Silero VAD, Whisper, llama.cpp, and Qwen3-TTS, all packed into 15G VRAM with hot-swapping capability.
View Cached Full Text
Cached at: 07/25/26, 10:13 PM
A cyber girlfriend made by a Bilibili bro, it totally feels right!
A purely locally running AI girlfriend, no internet connection, no API key.
VAD: Silero VAD v5
STT: Whisper
LLM: Local llama.cpp
TTS: Qwen3-TTS
All four models fit into 15GB VRAM and can be freely hot-swapped.
Modified based on the open-source project speech-to-speech.
When will my codex be able to do this? 😂 https://t.co/fre0BzJ9UH
Similar Articles
@laowangbabababa: Shocked! Dr. Qi on Douyin sells a 500k digital human agent per day, and I built it in 2 minutes. Using the Pixelle-Video project, which already has 22k stars. It supports digital human lip-syncing, motion transfer, and image-to-video. Supports ComfyUI, input a topic, from script writing to adding...
Introducing the open-source project Pixelle-Video: a fully automated AI short video engine. Input a topic and it automatically generates a video with script, images, voiceover, and background music. Supports local and cloud models, modular design allows flexible replacement of each component model.
@OpenBMB: Build with VoxCPM: Whispera — Your Local AI Voice Assistant What if your AI assistant could listen, think, remember, an…
Whispera 是一个基于 VoxCPM 的 Windows 本地实时语音助手，集成了 SenseVoice ASR、llama-server 本地 LLM 推理、VoxCPM 流式 TTS 和 Mem0 长期记忆，完全离线运行。
@FakeMaidenMaker: A lightweight personal AI agent that is small yet highly readable. Existing agent frameworks wrap the core loop in a black box; changing one detail means peeling through layers of abstraction, making even reading a chore, let alone secondary development. I'm recommending an ultra-lightweight open-source AI agent — nanobot. Designed for those who 'want to read and understand…'
Introducing the open-source ultra-lightweight AI agent nanobot. Its core loop is designed to be minimal and readable, making secondary development easy. It supports 10+ chat platforms including Telegram, Discord, Feishu, and multiple AI providers, with integrated long-term memory and MCP tool extensions.
@Soranlan: https://x.com/Bober_smart/status/2058223123720216767/video/1… An AI girl built for $20/month, then dual-platform on OnlyFans+Fanvue, TikTok breakout, Reddit precision targeting male fans, 7…
A 47-year-old Japanese man used the AI tool Claude to create an AI girl image for $20 per month, set up accounts on OnlyFans and Fanvue, and promoted on TikTok and Reddit, achieving 12 million views and $2,700 in revenue within a week, showcasing the potential of low-barrier AI content monetization.
@lxfater: NetEase Youdao open-sourced ZiYue 4 model, within 27B parameters, SOTA in math and science. But what really interests me is its voice feature!! Cloning a voice is nothing new, ElevenLabs could do it long ago. But they all share a common flaw: cross-language accent. Take your Chinese voice and use it to speak Japanese — it has a Chinese accent, you can tell it's a foreigner struggling...
NetEase Youdao open-sourced the ZiYue 4 model with 27B parameters, achieving SOTA in math and science; its voice feature supports 3-second cross-language voice cloning across 14 languages with no accent issue, along with open-sourcing the all-scenario intelligent agent 'Longxia' (Lobster).