惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

WordPress大学
WordPress大学
Cyberwarzone
Cyberwarzone
The GitHub Blog
The GitHub Blog
云风的 BLOG
云风的 BLOG
P
Proofpoint News Feed
小众软件
小众软件
Recent Announcements
Recent Announcements
博客园 - 三生石上(FineUI控件)
Security Archives - TechRepublic
Security Archives - TechRepublic
W
WeLiveSecurity
Cloudbric
Cloudbric
博客园 - 司徒正美
美团技术团队
N
News and Events Feed by Topic
Exploit-DB.com RSS Feed
Exploit-DB.com RSS Feed
PCI Perspectives
PCI Perspectives
宝玉的分享
宝玉的分享
H
Help Net Security
H
Hackread – Cybersecurity News, Data Breaches, AI and More
Google DeepMind News
Google DeepMind News
Help Net Security
Help Net Security
Last Week in AI
Last Week in AI
S
Schneier on Security
N
News | PayPal Newsroom
B
Blog RSS Feed
L
LINUX DO - 最新话题
T
Troy Hunt's Blog
S
Secure Thoughts
雷峰网
雷峰网
aimingoo的专栏
aimingoo的专栏
L
Lohrmann on Cybersecurity
G
Google Developers Blog
Microsoft Azure Blog
Microsoft Azure Blog
Hacker News - Newest:
Hacker News - Newest: "LLM"
T
Tenable Blog
S
Securelist
L
LangChain Blog
Recent Commits to openclaw:main
Recent Commits to openclaw:main
I
InfoQ
H
Heimdal Security Blog
Cisco Talos Blog
Cisco Talos Blog
F
Full Disclosure
Y
Y Combinator Blog
酷 壳 – CoolShell
酷 壳 – CoolShell
K
Kaspersky official blog
T
Tailwind CSS Blog
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
阮一峰的网络日志
阮一峰的网络日志
C
Cisco Blogs

Hacker News - Newest: "AI"

AI can't read an investor deck AI as an attorney? Student uses ChatGPT, Gemini to sue UW over alleged racial discrimination Hacking MCP Servers in AI Systems – The Rug Pull: Tool Changes After Approval GitHub - MeepCastana/KubeezCut: Free Web based video editor GitHub - GenAI-Gurus/awesome-eu-ai-act: Curated tools, official sources, OSS, templates, and guides for EU AI Act compliance. Can AI judge journalism? A Thiel-backed startup says yes, even if it risks chilling whistleblowers Coming soon: 10 Things That Matter in AI Right Now DARPA built an AI to fact-check enemy weapons claims What explains heterogeneity in AI adoption? When AI Meets Muscle: Context-Aware Electrical Stimulation Promises a New Way to Guide Human Movements - Department of Computer Science AI Changed How We Build. It Did Not Change What Matters. Linux rules on using AI-generated code - Copilot is OK, but humans must take 'full responsibility for the… Meta spins up AI version of Mark Zuckerberg to engage with employees Code Mode: Let Your AI Write Programs, Not Just Call Tools | TanStack Blog GitHub - Delavalom/graft: Go framework for building AI agents. Type-safe tools, multi-provider (OpenAI, Anthropic, Gemini, Bedrock), zero vendor SDKs. India's TCS tops estimates, says new AI models did not dent services demand Gen Z's fading AI hype Strong feeling: we are in a folded AI reality GitHub - machinarii/total-recall-catalog: A reference catalog of latest knowledge retrieval, memory & RAG systems GitHub - mensfeld/code-on-incus: Give each AI agent its own isolated machine with root, Docker, and systemd. Active defense detects and stops threats automatically.. Quantization, LoRA, and the 8% Problem: Benchmarking Local LLMs for Production AI Iran war: We spoke to the man making Lego-style AI videos that experts say are powerful propaganda Powell, Bessent discussed Anthropic's Mythos AI cyber threat with major U.S. banks GitHub - immartian/bellamem: Persistent belief-graph memory for AI agents. Retrieves decisive context by importance — not recency, not RAG, not /compact. recursive-mode: The Repo-Native Operating System for AI Engineering After the attack on Sam Altman's home, will AI CEO's go on the offensive? The biggest advance in AI since the LLM Opus 4.6 vs GPT 5.4 One Prompt Unity World Generation Test “AI polls” are fake polls Client Challenge Can AI be a 'child of God'? Inside Anthropic's meeting with Christian leaders How to Switch AI Chatbots and Why You Might Want To GitHub - MattMessinger1/agentic_refund_guardrail: Safe refund policy layer for AI agents — Python + TypeScript. Same behavior, shared tests. Adam/papers/emergent_values_whitepaper.md at master · strangeadvancedmarketing/Adam Ask HN: How do you stop playing 20 questions with your AI coding tools How far can automation and AI support psychotherapy? - @theU GitHub - stagas/rtdiff: realtime git diff gui and AI-assisted commits A Mac Studio for Local AI — 6 Months Later A History of the Early Years of AI at the University of Edinburgh Why AI Coding Tools Still Feel Stuck on Localhost MSN AI Datacenters Are Becoming Strategic Targets twitter.com Penn Researchers Use AI to Surface Unreported GLP-1 Side Effects in Reddit Posts Show HN: MoodSense AI (ML and FastAPI and Gradio, Deployed on Hugging Face) Moodsense Ai - a Hugging Face Space by aman179102 AI models are terrible at betting on soccer—especially xAI Grok GitHub - xialeistudio/echoic GitHub - HimashaHerath/github-dev-wrapped: AI-powered weekly GitHub activity reports deployed to GitHub Pages GitHub - alejandrobalderas/claude-code-from-source: Architecture, patterns & internals of Anthropic's AI coding agent — reverse-engineered from source maps AI and Tech brief: Ireland ascendant GitHub - Titovilal/context0: Context0 - Never Surrender Training for a Marathon with an AI Coach: What Worked and What Didn't Cyber Pulse: Agentic Intel - Apps on Google Play I Built an AI PR Reviewer That Catches Bugs by Not Looking for Bugs Gen Z workers are so fearful AI will take their job they’re intentionally sabotaging their company’s AI rollout | Fortune How AI Is Reimagining the Game of Golf–For Both Players and Courses GitHub - nattergabriel/reseed: A CLI tool for managing and distributing agent skills across projects Is SVG the final frontier? My AI workflow evolved from prompts to a near-autonomous workflow MLSharp Help - 3DGS Viewer & Generator I put my cognitive field based AI's runtime on GitHub Is Numble the first AI-proof game? A3: Kubernetes for autonomous AI agent fleets | Emergent Principles Deepali Vyas ("The Elite Recruiter") GitHub - msmarkgu/RelayFreeLLM: A restful API designed to route user prompts to various AI model providers. Unionized ProPublica staff are on strike over AI, layoffs, and wages Unleashing the Advantage of Quantum AI We're heading for an AI-fueled 'dementia crisis,' brain scientist warns The AI-Assisted Breach of Mexico's Government Infrastructure [pdf] GitHub - stef41/lmscan: 🔍 Detect AI-generated text and fingerprint which LLM wrote it. Open-source GPTZero alternative. Zero dependencies, works offline. MSN GitHub - visionscaper/collabmem: Enabling long-term collaboration with Agentic AI - building up episodic and world model memory over time with in-context awareness We gave an AI a 3 year retail lease in SF and asked it to make a profit | Andon Labs AI Code is Hollowing Out Open Source, and Maintainers are Looking the Other Way What leaked "SteamGPT" files could mean for the PC gaming platform's use of AI AI is the boss at this retail store. What could go wrong? GitHub - Wuzu11517/agentic-proxy: Local proxy meant to help reduce With Drones, Geophysics and ArtificiaI Intelligence, Researchers Prepare to Do Battle Against Land Mines A Single Operator, Two AI Platforms, Nine Government Agencies: The Full Technical Report 在 Steam 上购买 FriedrichAI: Offline AI 立省 10% GitHub - inevolin/resume-cli: Hit Claude usage limits? Resume any AI coding session elsewhere. Switch tools at zero friction. GitHub - atripati/ark: AI Runtime Kernel — a context operating system for AI agents. Eliminates tool bloat, loads only what’s needed, and gives LLMs their reasoning space back. How to Build a Secure AI PR Reviewer with Claude, GitHub Actions, and JavaScript This Startup Wants You to Pay Up to Talk With AI Versions of Human Experts Intel Arc Pro B70 Brings 32GB VRAM to Local AI for $949 WordPress 7.0: The Good, the AI, and the Still Missing AI on the couch: Anthropic gives Claude 20 hours of psychiatry IatroBench: Pre-Registered Evidence of Iatrogenic Harm from AI Safety Measures AI Agents Know About Supabase. They Don't Always Use It Right. The history and future of AI at Google, with Sundar Pichai Inside an AI‑enabled device code phishing campaign How Meta Used AI to Map Tribal Knowledge in Large-Scale Data Pipelines AI for Systems: Using LLMs to Optimize Database Query Execution Forecasting the Economic Effects of AI Introducing Tinker: Play with AI, bring your ideas to life AI sheds light on an ancient gaming mystery People really hate AI but not as much as Iran—or Democrats | Fortune What is an AI Product Engineer? Phoebe Gates wants her $185 million AI startup to succeed with 'no ties to my privilege or my last name': 'I have a chip on my shoulder' | Fortune
GitHub - typomonster/parlor-jarvis: On-device, real-time multimodal AI. Multilingual voice + vision (en/ko/es/pt/fr) with camera, screen, PDF, and video — runs entirely locally.
unusual_typo · 2026-04-27 · via Hacker News - Newest: "AI"

On-device, real-time multimodal AI. Have natural voice and vision conversations with an AI that runs entirely on your machine.

Parlor Jarvis is an enhanced fork of Parlor with stronger multilingual capabilities and richer image inputs. You talk, show your camera (or screen, PDF, or video), and it talks back — all locally.

Parlor Jarvis Demo

Video generation in ComfyUI.

Parlor Jarvis Demo

Image: AI-generated influencer

English translation of the demo conversation

User: I'm trying to generate a video using LTX, please help me.

AI: I see you're curious about generating a video using LTX. Currently, I can see a ComfyUI workflow and related documentation on the screen. Could you tell me more specifically what kind of video you'd like to make using LTX?

User: What clothes is the model wearing?

AI: The model in the photo is wearing a light green knit cardigan with a white top underneath. For the bottom, she is wearing denim jeans.

Parlor:

parlor_realtime_ai_with_audio_video_input_optimized.mp4

Research preview. This is an early experiment. Expect rough edges and bugs.

Parlor vs Parlor Jarvis

Capability Parlor Parlor Jarvis
Language model Gemma 4 E2B Supergemma 4 E4B (abliterated, enhanced Korean)
Text-to-speech Kokoro (English only) Supertonic — 5 languages: English (en), Korean (ko), Spanish (es), Portuguese (pt), French (fr)
Image input Camera Camera + screen sharing, PDF, video
Frontend Raw HTML Next.js

Why?

Local models are quickly catching up on both text generation and image understanding, and one of the most exciting breakthroughs is multilingual capability — strong performance across European languages (Spanish, Portuguese, French, …) and Asian languages (Korean, …), not just English. Parlor Jarvis is our take on what a fully on-device, multilingual, multimodal assistant can feel like today: speak any of the supported languages, point your camera, share your screen, drop in a PDF or video, and have a real conversation with a model running entirely on your machine. Enjoy!

Original author's why (from Parlor):

I'm self-hosting a totally free voice AI on my home server to help people learn speaking English. It has hundreds of monthly active users, and I've been thinking about how to keep it free while making it sustainable.

The obvious answer: run everything on-device, eliminating any server cost. Six months ago I needed an RTX 5090 to run just the voice models in real-time.

Google just released a super capable small model that I can run on my M3 Pro in real-time, with vision too! Sure you can't do agentic coding with this, but it is a game-changer for people learning a new language. Imagine a few years from now that people can run this locally on their phones. They can point their camera at objects and talk about them. And this model is multi-lingual, so people can always fallback to their native language if they want. This is essentially what OpenAI demoed a few years ago.

How it works

Browser (mic + camera / screen / PDF / video)
    │
    │  WebSocket (audio PCM + JPEG frames)
    ▼
FastAPI server
    ├── Supergemma 4 E4B via LiteRT-LM (GPU)  →  understands speech + vision
    └── Supertonic TTS (MLX on Mac, ONNX on Linux)  →  speaks back
    │
    │  WebSocket (streamed audio chunks)
    ▼
Browser (playback + transcript)
  • Voice Activity Detection in the browser (Silero VAD). Hands-free, no push-to-talk.
  • Barge-in. Interrupt the AI mid-sentence by speaking.
  • Sentence-level TTS streaming. Audio starts playing before the full response is generated.

Requirements

  • Python 3.12+
  • Node.js 20+ (for the Next.js frontend)
  • macOS with Apple Silicon, or Linux with a supported GPU
  • ~3 GB free RAM for the model

Quick start

Parlor Jarvis has two pieces: a FastAPI backend (src/) and a Next.js frontend (web/). In development you run both — Next proxies /ws to FastAPI so the browser talks to a single origin.

git clone https://github.com/typomonster/parlor-jarvis.git
cd parlor-jarvis

Terminal 1 — backend:

# Install uv if you don't have it
curl -LsSf https://astral.sh/uv/install.sh | sh

cd src
uv sync
uv run server.py

Terminal 2 — frontend:

cd web
npm install
npm run dev

Open http://localhost:3000, grant camera and microphone access, and start talking.

Models are downloaded automatically on first run (~2.6 GB for Gemma 4 E2B, plus TTS models).

How the dev proxy works

web/next.config.ts declares a rewrite from /wshttp://localhost:8000/ws, so the browser opens ws://localhost:3000/ws and Next.js proxies the upgrade through to FastAPI. Override the backend target with BACKEND_URL=http://host:port npm run dev if you want to point at a different server.

Configuration

Note: The default configuration is optimized for Korean, even though both the LLM and TTS support other languages natively. You can swap the LLM or TTS models via the env vars below — see Supported models and TTS engines.

Backend (src/)

Variable Default Description
MODEL_PATH auto-download from HuggingFace Path to a local .litertlm file
HF_REPO typomonster/supergemma4-e4b-abliterated-litert-lm HuggingFace repo to pull the model from
HF_FILENAME supergemma4-e4b-abliterated.litertlm File within HF_REPO to download
PORT 8000 FastAPI port
TTS_ENGINE supertonic supertonic (multilingual — en/ko/es/pt/fr) or kokoro (English, faster)
KOKORO_ONNX unset Force the Kokoro ONNX backend on Apple Silicon
SUPERTONIC_ONNX unset Force the Supertonic ONNX backend on Apple Silicon
SUPERTONIC_MLX_REPO typomonster/supertonic-2-mlx HF repo for the Supertonic MLX checkpoint
SUPERTONIC_ONNX_REPO Supertone/supertonic-2 HF repo for the Supertonic ONNX checkpoint

TTS engines

Parlor Jarvis ships with two interchangeable TTS backends, selected at runtime via TTS_ENGINE:

Engine Languages Apple Silicon Linux / x86
supertonic (default) en, ko, es, pt, fr mlx-audio fork1 (MLX) Supertonic ONNX Runtime
kokoro English only mlx-audio (MLX) kokoro-onnx

Force ONNX instead of MLX on macOS with SUPERTONIC_ONNX=1 or KOKORO_ONNX=1. Language is selected automatically from the frontend's active locale (Supertonic only — Kokoro ignores it). Set TTS_ENGINE=kokoro if you want the lighter English-only engine.

Supported models

Model HF_REPO HF_FILENAME
Supergemma4-E4B (abliterated) (default — strong multilingual, Korean) typomonster/supergemma4-e4b-abliterated-litert-lm supergemma4-e4b-abliterated.litertlm
Gemma 4 E2B (smallest, fastest) litert-community/gemma-4-E2B-it-litert-lm gemma-4-E2B-it.litertlm
Gemma 4 E4B (upstream base) litert-community/gemma-4-E4B-it-litert-lm gemma-4-E4B-it.litertlm

Switch models by setting both env vars, e.g.:

HF_REPO=litert-community/gemma-4-E2B-it-litert-lm \
HF_FILENAME=gemma-4-E2B-it.litertlm \
uv run server.py

Frontend (web/)

Variable Default Description
BACKEND_URL http://localhost:8000 Target for the /ws proxy in dev/build

Performance (Apple M3 Pro)

Stage Time
Speech + vision understanding ~1.8-2.2s
Response generation (~25 tokens) ~0.3s
Text-to-speech (1-3 sentences) ~0.3-0.7s
Total end-to-end ~2.5-3.0s

Decode speed: ~83 tokens/sec on GPU (Apple M3 Pro).

Project structure

parlor-jarvis/
├── src/                       # Python backend
│   ├── server.py              # FastAPI WebSocket server + Gemma 4 inference
│   ├── tts.py                 # Platform-aware TTS (MLX on Mac, ONNX on Linux)
│   ├── pyproject.toml         # Python dependencies
│   └── benchmarks/
│       ├── bench.py           # End-to-end WebSocket benchmark
│       └── benchmark_tts.py   # TTS backend comparison
└── web/                       # Next.js frontend (TypeScript + Tailwind + shadcn/ui)
    ├── app/
    │   ├── page.tsx           # Main UI — VAD, camera, audio playback
    │   ├── layout.tsx
    │   └── globals.css        # App styles + Tailwind tokens
    ├── components/ui/         # shadcn/ui components
    └── next.config.ts         # /ws rewrite → FastAPI

Acknowledgments

Our open source contributions

To make Parlor Jarvis possible, we ported and packaged a couple of upstream models for on-device use and shared everything back:

License

Apache 2.0

Footnotes

  1. mlx-audio @ git+https://github.com/typomonster/mlx-audio — the Supertonic-MLX support is pinned to this fork until it's upstreamed.