Voice-to-text for iOS and macOS, done well. Hotkeys, keyboard extension, formatting on the fly, meeting recording with AI summaries. More features than commercial apps that charge $100+/year, but free because one person built it in a few weeks with AI. Bring your own API keys or go fully offline with on-device engines if privacy matters to you.
Built-in dictation gives you raw text. SwiftSpeak formats it before it hits the page.
Raw text, manual cleanup
hey john just wanted to let you know that the project is delayed because we found some bugs in the code and we need to fix them before we can ship it to the client so the new deadline is going to be next friday instead of this wednesday hope thats okay let me know if you have any questions
No punctuation, no formatting, needs editing
AI-formatted, ready to send
Hi John,
Quick update on the project timeline: we discovered some bugs that need to be resolved before the client release. The new deadline is Friday instead of Wednesday.
Let me know if you have any questions.
Best regards
Professional email, ready to send
Show it an example of your writing and it matches your voice from then on. Different contexts for different apps.
Paste one example of how you write. The AI matches your style from that point on.
Gmail activates your Work context. WhatsApp activates Personal. Automatic, per app.
Three tiers: global, per-context, and per-session. The more you use it, the better it gets.
Examples & formatting chips
Memory & keyboard behavior
Voice-activated AI agents with memory, documents, and webhooks. Not templates.
Your Voice
Instructions
Your Docs
AI Memory
Webhooks
Output
Attach PDFs, research papers, notes. The AI reads them before answering, with citations.
Each Power Mode keeps its own memory. Your Research Assistant remembers your papers.
Send results to Slack, Notion, Zapier. Fetch context from APIs before processing.
Generate multiple versions. See the diff. Pick the best one. Refine with follow-ups.
Instructions & workflow memory
Documents (RAG) & webhooks
Draft professional emails from casual speech
Attach papers & docs, get cited answers
Describe changes, get formatted commits
Record meetings, get structured summaries
Voice your thoughts, get polished posts
Build your own voice-activated workflows
SwiftSpeak detects which app you're in and activates the right context automatically.
Gmail
Slack
Messenger
Teams
Notion
Figma
Zoom
X
Discord
Telegram
Gmail opens
→ "Work" context activates
Formal tone, business jargon
WhatsApp opens
→ "Personal" context activates
Casual tone, emoji allowed
Slack opens
→ "Technical" context activates
Code terms, concise style
iOS keyboard, macOS menu bar, system-wide hotkey. Dictate in any app, on any Apple device.
OpenAI, Anthropic, Google, Deepgram, AssemblyAI, DeepL, and more. Bring your own API keys.
Speak in any language, get text in another. Quality indicators show accuracy per language.
Medical, legal, technical, financial, scientific jargon. Add your own terms and acronyms.
Define your own formatting prompts. Email, casual, formal, or anything you need.
WhisperKit, Apple Intelligence, Apple Translation. Fully on-device, works offline.
See exactly what you spend. Daily, weekly, monthly breakdowns by provider.
Free. No subscriptions. No seat limits. Use it, fork it, build on it.
Use cloud providers for best accuracy, or keep everything local for privacy. Mix and match.
Best accuracy, your API keys
Fully private, works offline
I built SwiftSpeak in a few weeks, one person with AI coding agents. It does what apps charging $100-200/year do. If that's possible, the subscription model for this category of software doesn't have long. I'd rather share it.
Wispr Flow
$12-15/mo
$144-180/year
Otter.ai Pro
$8.33/mo
$100/year, 1200 min cap
Otter.ai Business
$20-30/mo
$240-360/year
SwiftSpeak
$0
open source, forever
You bring your own API keys. These are the real costs from the providers.
Light use
50 short transcriptions/day
~$0.15/mo
Moderate use
150 transcriptions/day
~$0.45/mo
Heavy use
400 longer transcriptions/day
~$1.20/mo
Or use WhisperKit + Apple Intelligence for $0. Fully local, no API keys needed.
Total spend & trends
Cost by provider & category
Free and open source. No account required, no subscriptions.
Universal (Intel + Apple Silicon)
Requires macOS 13.5+
Download .zipFirst launch: right-click the app → Open
Already using DictAll for macOS? SwiftSpeak builds on the same ideas.
SwiftSpeak is built by Paweł Gawliczek