www.transcriptionists.com

Show HN: OneSentence – An offline macOS voice utility built entirely with AI

Hi HN, I’m sharing OneSentence, an offline voice utility for macOS (M-series). I built this for two reasons: first, I wanted to see how far I could push cheap AI, and second, I wanted to use this utility. The idea was born out of using Emacs packages with Whisper to dictate to my machine. I had found it effective to simply speak and articulate context to coding agents. OneSentence does four things well: privacy, speech-to-text, text-to-speech, and template insertion.The development process was p

Show HN: Videolyti – Free video downloader with built-in AI transcription

I got tired of juggling three or four different sites every time I needed to download a video and grab the transcript. TikTok downloaders are plastered with fake buttons. YouTube converters redirect you through five pages. And actual transcription costs money.So I built Videolyti over a few months. You paste a URL from YouTube, TikTok, Instagram, Twitter, Facebook, Reddit, or Vimeo — it gives you the video file and a text transcript.The transcription runs OpenAI Whisper (large-v3) on my own serv

Show HN: OneCamp – Self-Hosted Slack/Asana/Zoom/Notion Alternative

Launching in 6 days (March 7)!OneCamp is a self-hosted unified workspace that combines real-time chat, tasks, video calls, and collaborative docs — no per-user fees, unlimited users, full data control.We open-sourced the entire Next.js frontend so anyone can explore, fork, or contribute:https://github.com/OneMana-Soft/OneCamp-feKey pieces of the architecture:1. Real-time collaboration - Yjs + Hocuspocus (CRDT sync over WebSockets) - Tiptap editor + custom Node micro

Show HN: Yakki.ai – Say it. Ship it. Increase your output with your voice

I built a macOS menu bar app for capturing and leveraging the audio you produce/generate.Hold fn, talk, release, your text appears wherever your cursor is. Transcription runs on-device using local models, works offline, and nothing leaves your machine if you do not want to.I wanted a faster dictation app that was put me in control, but also let me do things like recording my meetings without having to invite any weird guess to the Teams call, generate perfect meeting notes with insights, an

Show HN: Axiom – structural OCR for handwritten STEM notes

I built Axiom after repeatedly running into the same problem with my own handwritten STEM notes.On paper, everything looks clean — equations aligned, steps grouped properly, tables laid out clearly. But the moment I scanned those pages and ran them through OCR (including LLM-based tools), the structure would fall apart. The characters were mostly correct, but the layout — which is what actually makes math readable — was gone.Aligned equations would lose alignment. Multi-step derivations would co

Show HN: I built a sub-500ms latency voice agent from scratch

I built a voice agent from scratch that averages ~400ms end-to-end latency (phone stop → first syllable). That’s with full STT → LLM → TTS in the loop, clean barge-ins, and no precomputed responses.What moved the needle:Voice is a turn-taking problem, not a transcription problem. VAD alone fails; you need semantic end-of-turn detection.The system reduces to one loop: speaking vs listening. The two transitions - cancel instantly on barge-in, respond instantly on end-of-turn - define the experienc

Google just killed my project

For the past year, I’ve been building GM Pro — a Chrome extension that upgrades the chat experience inside Google Meet.It started simple: reactions, replies, mentions, dark mode for chat. Then I added auto-join, auto-mute, transcription tools, lobby notifications, attendee shuffling. Basically all the things you wish Meet chat had by default.People loved it. 5-star reviews. Steady installs. Real usage.And then, after many years of lackluster chat, Google announced they’re integrating Meet chat d

Show HN: Video to Text AI Transcription

I’ve been building a video-to-text web app and wanted to share it for feedback. The core flow is straightforward: upload files, start transcription, then track progress in a history page that refreshes automatically while jobs are running. Paid users can submit multiple files at once, and speaker diarization is supported for conversations and interviews.Over the last few weeks I focused mostly on reliability. I changed the pipeline to extract audio first and then run transcription, which made lo

Aura-State: Formally Verified LLM State Machine Compiler

I noticed a pattern: every LLM framework today lets the AI manage state and do math. Then we wonder why pipelines hallucinate numbers and break at 3 AM.I took a different approach and built Aura-State, an open-source Python framework that compiles LLM workflows into formally verified state machines.Instead of hoping the AI figures it out, I brought in real algorithms from hardware verification and statistical learning:CTL Model Checking: the same technique used to verify flight control systems,

Show HN: BananaOS, vibecoded operating system that boots on a 486 with ~11MB RAM

My 10-year-old son has been deep in low-level rabbit holes lately and ended up vibe-coding his own operating system. Since he’s still a kid and not on HN himself, I’m posting this on his behalf with his permission.This started as curiosity about how computers actually boot, and somehow escalated into writing a kernel, building a GUI, and setting up CI that produces a bootable OS image on every commit.BananaOS is a small experimental operating system built mainly for learning and exploration of l

Is traditional ML relevant anymore? Any active research going on in ML methods?

Everything I see today is only about LLMs. I personally use LLMs in my daily activities and they are doing great job. But I wonder what happened to traditional machine learning methods all of a sudden! Those hamspam classifiers, sentiment analysis models, word2vecs, RNNs, CNNs, LSTMs, FFNNs, where are they now? What is a typical data scientist or a ML engineer of late 2010s doing now? The decision trees they trained, the neural nets they architected, the accuracy evaluations, hyperparameter tuni

Show HN: CosmicMeta – Daily AI and tech analysis with a humanization pipeline

I built CosmicMeta.ai, a tech platform that publishes daily analysis on AI, machine learning, and emerging tech.The interesting technical bit: every article goes through a two-pass humanization step that detects and rewrites 24 specific AI writing patterns (significance inflation, copula avoidance like "serves as" instead of "is", em-dash overuse, formulaic conclusions, etc.). It's based on research from the blader/humanizer framework.Tech stack: Spring Boot, OpenAI

Spotify's take on ADRs is great, but how do you enforce them at scale?

Hey HN,I built Decision Guardian — an open-source GitHub Action and CLI that automatically surfaces architectural decisions as PR comments when code touches protected files. The problem it solves:Spotify published a great post in 2020 about when to write Architecture Decision Records. I followed the advice. My team wrote ADRs. They sat in docs/adr/. Nobody read them before opening a PR.https://engineering.atspotify.com/2020/04/when-should-i-write-an-architectur

MRIs Reveal Brain Structure Changes After Second Pregnancy

WOMEN experience unique changes in grey matter volume, white matter tracts, and functional neural network organisation after ...

Ellucian Returns as Sponsor of 2026 HBCU AI Conference and Training Summit to Support Community-Centered AI Innovation and Leadership

Ellucian will serve as a Neural Network Sponsor of the 2026 HBCU AI Conference and Training Summit, March 10–11, 2026, at Huston-Tillotson University in Austin, Texas. Ellucian will host a mainstage ...

Google’s newest AI agents bring telcos a step closer to autonomous network operations

Google LLC is getting closer to fulfilling its vision of enabling truly autonomous network operations with the launch of its ...

Chinese researchers create neural network for modeling human concept formation

BEIJING -- Chinese scientists have developed a novel neural network that enables artificial intelligence (AI) to form ...

Following Up On President Trump’s Idea Of Renaming AI

Forbes contributors publish independent expert analyses and insights. Dr. Lance B. Eliot is a world-renowned AI scientist and consultant. President Trump included a seemingly ad hoc remark during his ...

Artificial intelligence could reduce emergency room mental health visits at CHEO

Researchers at CHEO have enlisted artificial intelligence to help with a difficult problem: the high number of children and ...

Researchers say artificial intelligence is being used in swatting attacks

Officials said swatting is often used to disrupt and cause panic in communities. Now, researchers said the people committing the crime are utilizing new technologies, like AI.