Voice & Multimodal AI

A Rust and Tauri desktop app that combines capture-proof window tricks, offline voice tracking, and remote script collaboration into one lightweight teleprompter.

How a minimalist Swift utility uses computer vision to kill the "Cmd+Tab" bottleneck in multi-monitor workflows.

A look at the schema-first design that powers tailored questions, prep roadmaps, and ATS-friendly resume PDFs without letting the model run wild.

How a zero-unsafe Rust orchestrator unified the fragmented Whisper ecosystem into a high-reliability streaming engine for AI agents.

A local-first dictation and assistant app for Windows, built around Whisper, Ollama, and a clipboard-safe way to paste speech into any focused app.

A tiny SwiftUI utility that starts recording on launch, runs the audio through Whisper, and posts the result into Notion with almost no interface in between.

A protocol-first MCP server that maps OBS WebSocket v5 into safe, structured tools, so an AI assistant can switch scenes, mute sources, and drive a live production without falling apart when OBS does.

How an intrinsic speaker encoder and Flow-VAE synthesis jumped to the top of the TTS Arena leaderboard.

A deep look at the local-first overlay that uses NSPanel, global event taps, and live screen context to make AI feel like a system reflex instead of a separate app.

A guided installer, custom DSP scripts, and game-specific tuning files turn messy Windows audio into a tactical system for competitive play.

A production-minded AI phone framework that mixes LiveKit, Sarvam AI, SIP telephony, and a browser control panel to automate booking, support, and lead qualification in Hinglish and regional languages.

A tiny Manifest V3 wrapper, a configurable iframe, and a maxed-out z-index turn one Dify app into a browser-wide assistant. The interesting work here is not intelligence. It is delivery.

A universal AI sidebar is the visible feature. The real invention is a privacy-first prompt triage and learning loop that lives entirely inside the browser.

A real-time audio-visual model that uses mouth motion, causal convolutions, and SRUs to pull one voice out of a crowd without peeking into the future.

Google Research’s open-vocabulary video segmentation stack pairs language grounding, memory, and deployment-minded engineering so the hard part is not just accuracy, but speed.

A causal separator from Tsinghua that swaps full hindsight for a rolling memory, and tries to erase the usual quality penalty of live audio.

A bridge between Even G2 smart glasses and a terminal agent, with voice input, approval routing, and privacy-aware state handling built from surprisingly simple pieces.

A self-hosted collaboration stack that uses proximity, WebRTC, and a browser game engine to make remote work feel spatial, persistent, and strangely tangible.

A tiny `UIViewRepresentable`, a cache, and one clever request guard make animated characters behave like app state, not flaky media assets.

How a synthetic show becomes a seamless channel by trimming dead air, looping clips, and broadcasting through Replicate, Fly.io, and Cloudflare.

Apollo-data-preprocess is not a model repo. It is the filter that strips silence, slices tracks into learnable windows, and turns raw stems into HDF5 training fuel for high-fidelity music restoration.

How an offline-first voice assistant uses Apple's MLX framework and aggressive audio engineering to sever the cloud API umbilical cord.

A performance-obsessed take on generative video that favors instant steering, multimodal input, and rough-edged feedback over polished cinematic output.

A thin Cloudflare Worker keeps the keys safe, WebRTC keeps the voice loop fast, and client-side tool calls turn speech into immediate action.

A tiny Swift utility that hides UIKit boilerplate, pre-warms the Taptic Engine, and turns haptics into a readable intent call.

This repo is the training scaffolding behind Gull, a generative audio codec that compresses and reconstructs sound from subbands instead of raw waveforms.

A small OpenCode plugin uses ElevenLabs v3, expressive audio tags, and macOS playback to make an AI agent sound directed, not robotic.

A C++ Docker sample that keeps the SDK alive, authenticates on the fly, and routes raw meeting audio into recording or transcription pipelines.

How a lightweight macOS utility hijacks the Right Option key and uses AppleScript to give AI agents eyes on your active window.

A curated leaderboard that ranks the audio field three ways at once, and reveals why the strongest projects now win on research, demo quality, and open-source execution together.

A spatial voice assistant that fuses live transcription, multi-monitor vision, and cursor control into one unusually teachable desktop workflow.

A real-time audio generator that trades loops and media players for a minimal, low-GC pipeline, a first-order filter, and a shell-script switch you can actually live with.

A local-first Chrome extension that watches live captions, prioritizes what you can see, falls back across LLMs when a model fails, and keeps the whole meeting history on your device.

How a local-first workflow uses multimodal LLMs to solve the tedious math of lyric alignment and AI lip-syncing.

A deep look at the parser pipeline, signal detection, and voice layer behind an open-source soundbox alternative for Android merchants.

A hybrid biometrics prototype combines Gabor filtering, handcrafted texture features, EfficientNet, and heuristic heatmaps to estimate ABO/Rh blood groups from fingerprints.

A real-time Pictionary-style stack where the server owns the room, the canvas stays aligned across devices, and offline players get skipped instead of breaking the match.

A Next.js dashboard that listens, classifies, and executes. The real trick is not the interface, but the translation layer between human speech and structured action.

Clerk handles identity, Convex keeps the room in sync, and Stream carries the call. The interesting part is how cleanly the handoffs work.

A drop-in macOS speech command that routes text to cloud TTS, local models, warm daemons, and fallback voices without changing your scripts.

Breaking the terminal tether with asynchronous voice handoffs and persistent memory loops.

A deep look at how this repo combines structured prompt engineering, browser speech APIs, and credit-backed SaaS logic to simulate a realistic mock interview.

A real-time voice interview stack that treats Gemini like a component, not a decision-maker.

A local-first pipeline for transcription, translation, and burn-in subtitles, with the agent doing the editorial work where it matters most.

A voice-first mock interview app uses Gemini, Web Speech, and Firebase to turn a simple frontend into a tense, cinematic coaching loop that survives failed mic input, slow models, and empty transcripts.

A deep dive into the verified-provider gate, credit ledger, and just-in-time video provisioning that make this Next.js platform feel more like a controlled marketplace than a simple booking tool.

A lightweight Flask project for visualizing clinical trial data, built with a compact front end, a small contributor surface, and just enough structure to keep the workflow legible.

A Streamlit app that first describes an image literally, then hands that description to an LLM for captions and hashtags. The interesting part is not the UI. It is the chain.