Hi HN,

I built a voice-to-text tool that runs entirely in your browser. No account required for the free tier, no data sent to my servers.

Try it: https://voicetotextonline.com

Why I built this:

- Existing tools require signups, have minute limits, or cost money - Google Docs voice typing requires a Google account - Dragon costs $150-500 - Otter.ai has free tier limits

(A) Free Features (no account required):

1/ Core Transcription:

- Real-time voice-to-text using Web Speech API - 55+ languages supported - Auto-punctuation & sentence case options - Works offline after first load (PWA)

2/ AI Enhance (added based on user survey – 80% voted yes):

- Auto-fix grammar, punctuation & formatting - One-click cleanup of transcripts

3/ My Projects (local storage):

- Save transcripts to browser localStorage - Organize with folders (Notes, Work, Personal, etc.) - Custom folders & tags - Search across all transcripts - Edit, copy, download as TXT - 100% private – never leaves your device

- Export:

- Copy to clipboard - Download as TXT or DOCX

(B) Pro Features ($10/month or $1/hour pay-per-use):

1/ File Upload & Transcription:

- Upload audio/video files (MP3, WAV, M4A, MP4, MOV, AVI, MKV) - Up to 500MB per file - Batch upload (10 files at once) - Powered by AssemblyAI (95%+ accuracy) - 150 hours/month transcription

2/ Advanced Features:

- Real-time progress with ETA - Speaker labels - In-browser audio recording (5 min with pause/resume) - Translation to 25+ languages (GPT-4o)

3/ Export Formats:

- TXT, SRT, VTT, JSON with timestamps - Segment-level timestamp precision

4/ Cloud Storage:

- Transcription history in the cloud - 10 GB storage, 1,000 files/month

Free tier:

- All transcripts stored in browser localStorage only - Never touches our servers - 100% private

Pro tier:

- Audio files stored in Supabase (encrypted) - Files retained for 30 days for re-download, then auto-deleted - Transcripts stored permanently in your account - You can delete any transcript or your entire account anytime - We don't use your data for training

Tech stack:

- Next.js 14 (App Router) - Web Speech API (free real-time transcription) - AssemblyAI (Pro file transcription, 95%+ accuracy) - OpenAI GPT-4o (AI Enhance & translation) - Supabase (auth & storage) - Stripe (payments) - Tailwind CSS - Hosted on Vercel

Limitations:

- Real-time transcription doesn't work in Firefox (Web Speech API not supported) - Free tier accuracy depends on Chrome's speech engine

Would love feedback on UX, pricing, or feature ideas. Considering open-sourcing the core transcription component.

Jeffrey Way: I'm Done

Ask HN

Fast Joystick Read – Elon Musk on Usenet, 1994

World’s most powerful literary critic is on TikTok

Non-Functional Requirements: The Secret Sauce Nobody Wants to Season

Show HN: Voice to Text– Free browser-based speech-to-text with local projects

Show HN: Structured data extraction using local quantized LLMs

Jurassic Park: SGI Computers (2010)

Earth's Rotation Limits IBIS Performance to 6.3 Stops

Lawsuit Claims Meta Can See WhatsApp Chats in Breach of Privacy

The Brain of the Greatest Solo Climber

'Jurassic Park' at 30: The Gift That Keeps on Giving (2023)

Jurassic Park: Effects Team Brings Dinosaurs Back from Extinction (1993)

Oculomics and AI: The eye as a biomarker for health span

New AI Architecture

Show HN: PodCost – Find wasted GPU and Kubernetes spend (with live demo)

Compiling Models to Megakernels

Vnsh – An ephemeral, host-blind file sharing tool for AI context

What Ralph Wiggum loops are missing

AI won't take your job. A guy with a $599 Mac Mini and Claude will

149M Infostealer Credentials Exposed in Unsecured Database

What If We Took Message-Passing Seriously?

Because Coordination Is Expensive

Palantir has no place in UK public services

Canon and Compass into <Si Wu Zi> Scripture

Eightfold AI sued for job candidate reports without their consent

Debugging a "weird SIGSEGV" core dump with Codex and GDB inside Docker

The Great Escape and Stalag Luft III

Trump Family's $6.8B Fortune Is Increasingly Tied to Crypto

Stable-DiffCoder: Pushing the Frontier of Code Diffusion Large Language Models

Ask HN: A good Model to choose in Ollama to run on Claude Code