{"slug": "show-hn-dicty-privacy-first-local-voice-dictation-and-dev-documentation-tool", "title": "Show HN: Dicty – Privacy-first local voice dictation and dev documentation tool", "summary": "Dicty launched as a privacy-first local voice dictation and developer documentation tool for macOS and Windows, offering offline speech recognition through Whisper Nano, Base, Fast, Turbo, and Pro models plus NVIDIA Parakeet, with zero-latency transcription in 90+ languages. The app supports 400+ local and cloud language models, including offline GGUFs such as Qwen, Llama, Mistral, and DeepSeek R1 and cloud options like Claude 3.7, DeepSeek V3, GPT-5, Gemini, and Grok via user-supplied API keys, and its core engine is free with no subscriptions, minute limits, or audio tracking. Dicty syncs generated documentation to Obsidian, Linear, Notion, and Confluence, and its Pro tier adds offline transcription of .mp3, .m4a, and .wav audio and video files.", "body_md": "# Dictate.Transcribe.Transform.\n\nInstant voice-to-text for macOS and Windows. Dictate anywhere, format with any LLM, and sync directly with your Git workflow.\n\n## Engineered for control, privacy, and speed.\n\nSee the actual desktop interface. No hidden cloud lock-in—manage your local speech models, custom prompt modes, and API keys with complete transparency.\n\n### Offline Speech Models with Full Benchmarks\n\nDicty gives you direct control over your local speech recognition engine. Download and switch between Whisper Nano, Base, Fast, Turbo, and Pro models directly within the desktop app.\n\nMaximum Modularity\n\n## Modular, private, and customizable.\n\nChoose your speech recognizer, your LLM refiner, and your formatting logic with zero vendor lock-in.\n\n### Fast Speech Recognition & Live Translation\n\nZero-latency local transcription in 90+ languages. Choose from Whisper (Nano to Ultra Turbo) or NVIDIA Parakeet models without cloud delay.\n\n### 400+ Local & Cloud Language Models\n\nRun offline GGUFs locally (Qwen, Llama, Mistral, DeepSeek R1) or connect cloud LLMs like Claude 3.7, DeepSeek V3, GPT-5, Gemini, and Grok with your own API keys.\n\n### Smart Formatting & Custom Prompts\n\nTransform raw speech into clean prose, commit notes, or structured meeting summaries. Customize prompts directly in the app.\n\n### 100% Private & Free Core Engine\n\nNo subscriptions, minute limits, or audio tracking. Your voice and transcripts stay completely on your machine.\n\nThe Architecture\n\n## Built for how engineers actually think.\n\nFive core systems that turn scattered spoken intent into durable, connected knowledge, or handle your everyday transcriptions flawlessly.\n\n### Rapid Voice Capture\n\nPush hotkey, speak your thoughts, done. Captures the 'Why' behind code changes without losing flow or breaking context.\n\n### Link Directly to Code\n\nGround your docs in reality. Select related PRs, commits, and branches so every doc links directly to production code.\n\n### Living, Self-Updating Docs\n\nPrevents documentation decay. Reconciles new updates with existing architecture files rather than creating redundant, linear logs.\n\n### Sync to Any Tool or Vault\n\nYour docs, your vault. Seamless native export to Obsidian, Linear, Notion, or Confluence.\n\n### Audio & Lecture Transcription (Pro)\n\nNever write down lectures or interviews by hand again. With Dicty Pro, drop any recorded audio (.mp3, .m4a, .wav) or video into Dicty for instant offline transcription, then let local or cloud AI summarize key takeaways, formulas, and study notes without time limits.\n\n### Translate Text to Voice on the Fly\n\nHighlight text anywhere—documentation, PR diffs, articles, or foreign copy—to translate text to voice on the fly. Powered by dual local neural engines (Kokoro-82M and Silero 48kHz) for studio-grade, real-time speech without cloud latency or clipboard overwrite.\n\nThe Architectural Librarian\n\n## Stop making messy notes that nobody reads.\n\nOther tools just pile up random notes that get lost forever. Dicty is different. You speak, and it puts your thoughts right where they belong—clean, organized, and always in one place.\n\n### Linear logging\n\nAppend-only. Redundant. Decaying.\n\nFour documents describe the same system. Which one is true? Nobody knows.\n\n### Living architecture docs\n\nSelf-healing. Reconciled. Authoritative.\n\nEvery input (voice notes, PRs, standups, commits) feeds directly into your central **ARCHITECTURE.md**. One canonical source of truth, always current, always grounded in code.\n\nHow it Works\n\n## From spoken word to synced doc in four steps.\n\n### Speak\n\nTrigger via local shortcut and dictate your architectural intent or task context.\n\n### Select Artifacts\n\nAttach relevant GitHub PRs, commits, or ticket links.\n\n### Reconcile\n\nDicty's AI synthesizes the diffs and voice input into structured markdown.\n\n### Direct Sync\n\nPush updated docs straight to your target knowledge storage.\n\nIntegrations & Ecosystem\n\n## Works seamlessly with your tech stack, local AI engines, and team vaults.\n\n## How Dicty compares to alternative apps.\n\nEngineered specifically for developers, architects, and privacy-conscious professionals. Zero subscription paywalls.\n\n| Capability | Dicty.io | Superwhisper | MacWhisper | Wispr Flow | \n|---|---|---|---|---|\n| 100% Free Desktop Engine (No Monthly Subscriptions) |  | $8.49/mo | €64 Pro | €15/mo | \n| Cross-Platform Availability (macOS & Windows) |  | Mac Only | Mac Only |  | \n| Zero-Latency Hardware Acceleration (Metal on Mac, AVX2 on Windows) |  | Metal Only | Metal Only |  | \n| Choice of Speech Models (NVIDIA Parakeet, Whisper Ultra) |  |  | Whisper Only |  | \n| Active Codebase Context (Git PRs, Commits, Branches) |  |  |  |  | \n| Architectural Librarian & Multi-Vault Sync (Obsidian, GitHub, Linear) |  |  |  |  | \n| 400+ Cloud & Local Text Models (Claude, GPT-5, Gemini, Grok, DeepSeek) |  | Limited |  |  | \n| Complete Privacy (Zero Audio Recorded to External Cloud) |  |  |  |  | \n\nLooking for granular feature breakdowns, pricing math, and architectural details?\n\n[View All Side-by-Side Comparisons](https://dicty.io/compare)\n\n## 100% free on your Mac & Windows. Upgrade only for extra cloud AI.\n\nSpeak as much as you want for $0. No limits, no subscriptions. Upgrade to Pro only if you want built-in cloud models and automatic cloud sync.\n\n### Starter\n\nForever Free & Private\n\n- 100% local speech engine (Whisper, NVIDIA Parakeet)\n- Unlimited offline voice dictation & typing\n- Architectural Librarian & Git branch context\n- Speaker Diarization (Up to 2 hrs / session)\n- Up to 3 custom system prompts & styles\n- Up to 20 custom vocabulary terms & acronyms\n- BYOK (Bring-Your-Own-Key) for OpenRouter & Claude\n- Built-in Cloud AI Quota: Inactive (BYOK only)\n\n[Download App](https://dicty.io/download)\n\n### Pro\n\n- Everything in Starter\n- Active Built-in Cloud AI Quota (Gemini 2.0, Claude 3.5, DeepSeek, GPT-4o)\n- Unlimited Custom System Prompts & Modes\n- Unlimited Vocabulary & Custom Acronyms\n- Multi-Vault Cloud Sync (Obsidian, Notion, Confluence, GitHub)\n- Batch Audio & Video File Transcription + Diarization\n- Multi-Project Profiles & Fast Switching\n- Priority Model Routing & Early Beta Access\n\n[Get Pro](https://dicty-backend.dicty-backend.workers.dev/auth/google?source=web)\n\n### Early Supporter\n\nLifetime Community License\n\n- All Pro features unlocked forever\n- Grandfathered benefits & cloud compute balance\n- Direct roadmap voting & priority feedback\n- Private developer channel access (VIP Discord)\n- Founding supporter badge & zero recurring fees\n- Support independent developer tooling\n\n[Buy Early Supporter](https://dicty-backend.dicty-backend.workers.dev/auth/google?source=web)\n\n### Detailed Comparison Matrix\n\nFull breakdown of limits, audio processing engines, and local vs. cloud features.\n\n| Feature | Starter ($0) | Pro ($4/mo) | Early Supporter ($99) | \n|---|---|---|---|\n| Speech & Audio Engine |  |  |  | \n| On-Device Whisper (C++ / Metal & AVX2) Metal GPU acceleration on Mac and AVX2 CPU on Windows (Nano to Large V3 Turbo) | Unlimited | Unlimited | Unlimited | \n| NVIDIA Parakeet (TDT / CTC) Sub-400ms low latency speech model | Unlimited | Unlimited | Unlimited | \n| Translate Text to Voice on the Fly Instant speech synthesis on highlighted text. Kokoro-82M vector blending & Silero 48kHz broadcast speech | Unlimited | Unlimited | Unlimited | \n| Audio Privacy & Zero-Retention Audio processed in RAM; never leaves your machine |  |  |  | \n| Meeting Audio Duration Microphone and system loopback recording | Up to 2 hrs / session | Unlimited | Unlimited | \n| Speaker Diarization & Dialogue Scripting Multi-speaker turn separation & executive summaries |  |  |  | \n| Audio & Video File Transcription Batch drag & drop for lectures, meetings, and interviews (.mp3, .m4a, .wav, .mp4) |  |  |  | \n| AI Reasoning & Models |  |  |  | \n| Bring-Your-Own-Key (BYOK) Connect your personal OpenRouter, OpenAI, Claude, or local Ollama keys | Unlimited | Unlimited | Unlimited | \n| Built-in Managed Cloud AI Quota Zero setup Gemini 2.0 Flash, Claude 3.5 Haiku, DeepSeek V3, and GPT-4o tokens without API keys | Inactive (BYOK only) | Included monthly quota | Included + Top-ups | \n| Custom System Prompts & Modes Tailored formatting styles (Email, Code, Slack, Jira, Book Prose) | Up to 3 presets | Unlimited | Unlimited | \n| Custom Vocabulary & Acronym Library Specialized tech stack jargon, libraries, and name phonetic anchors | Up to 20 words | Unlimited | Unlimited | \n| Architectural Librarian & Context |  |  |  | \n| Architectural Librarian (VKE) Voice-to-Knowledge Engine synthesizing architecture documentation |  |  |  | \n| Git Provider Context (PRs, Commits, Branches) Inspects diffs and active PRs across GitHub, GitLab, and Bitbucket | Local & Active Branch | Full Multi-Repo Context | Full Multi-Repo Context | \n| Issue Tracker Context (Linear, Jira) Links spoken insights to existing tickets and backlog items | Manual linking | Auto-sync & Fetching | Auto-sync & Fetching | \n| Multi-Vault Sync (Obsidian, Notion, Confluence) Automated documentation commit and push to remote wikis | Local export | Direct Remote Sync | Direct Remote Sync | \n| Workspaces & Access |  |  |  | \n| Project & Workspace Profiles Custom setups per client or repository with isolated prompts | 1 Active Profile | Unlimited Profiles | Unlimited Profiles | \n| License Term Subscription billing frequency | Free Forever | Monthly or Yearly | Lifetime Access | \n| Roadmap Voting & Community Channel Direct influence on feature development and VIP Discord channel | [Community](https://discord.gg/CNEkeVj3b) | Priority Support | Direct Roadmap Voting & VIP Channel | \n\nLooking to equip your entire engineering team? [Contact us for Team & Enterprise Deployments](https://dicty.io/cdn-cgi/l/email-protection#1e767b727271307a777d6a673077715e79737f7772307d7173)\n\n[Have questions before choosing a plan? Discuss with users and the team on Discord](https://discord.gg/CNEkeVj3b)\n\nGot Questions?\n\n## Frequently Asked Questions\n\nEverything you need to know about Dicty’s privacy, offline capabilities, AI customization, and performance.\n\nYes! The core desktop app, local Whisper speech models (from ultra-fast Nano to studio-grade Large V3 Turbo), local audio file transcribers, and offline LLM integrations (like LM Studio and Ollama) are 100% free. There are no monthly paywalls, no artificial audio duration cut-offs, and no hidden subscriptions.", "url": "https://wpnews.pro/news/show-hn-dicty-privacy-first-local-voice-dictation-and-dev-documentation-tool", "canonical_source": "https://dicty.io", "published_at": "2026-09-27 12:30:58+00:00", "updated_at": "2026-09-27 13:01:41.251861+00:00", "lang": "en", "topics": ["ai-tools", "ai-products", "natural-language-processing", "large-language-models", "developer-tools"], "entities": ["Dicty", "Whisper", "NVIDIA Parakeet", "Qwen", "Llama", "Mistral", "DeepSeek R1", "Claude 3.7"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/show-hn-dicty-privacy-first-local-voice-dictation-and-dev-documentation-tool", "markdown": "https://wpnews.pro/news/show-hn-dicty-privacy-first-local-voice-dictation-and-dev-documentation-tool.md", "text": "https://wpnews.pro/news/show-hn-dicty-privacy-first-local-voice-dictation-and-dev-documentation-tool.txt", "jsonld": "https://wpnews.pro/news/show-hn-dicty-privacy-first-local-voice-dictation-and-dev-documentation-tool.jsonld"}}