Disclosure: Paraspeech is our product. It’s included here with the same standard as everything else: concrete use cases, clear limits, and sources.
This list is intentionally tool-first: things you install or subscribe to because they directly reduce keystrokes, context switching, or time-to-feedback.
1) Paraspeech (macOS) - Mac-first dictation for developers
Paraspeech is a menu bar voice-to-text app for Mac. You hold a hotkey, speak, and release; with Accessibility permission, it can insert text into many editable Mac fields, while unsupported fields may require a clipboard fallback. When the active speech backend is local, speech audio is transcribed on the Mac. Rewrite uses a separate backend, and eligible cloud speech—including allowed temporary fallback—or cloud rewrite transmits the content needed for that operation.
What makes it different (and why devs care)
- Mac-wide input: can insert into many editable fields in IDEs, terminals, browsers, PR comment boxes, Slack, Jira, and Notion; test your exact field and clipboard fallback.
- Local and cloud-backed paths: supported local modes are available on Apple Silicon, while eligible cloud paths and allowed temporary speech fallback require internet.
- Menu bar workflow: designed to stay available from the Mac menu bar.
- Developer-friendly controls: hotkey modes (hold / toggle), Auto-Send vs clipboard-only, history, privacy mode, quick cancel.
- AI Rewriting: use Cloud Cleanup or, where the direct Mac distribution and current product access permit it, an on-device rewrite path for emails, chats, notes, and other everyday writing.
Quick facts
- Mac support: Universal Mac app for Intel and Apple Silicon Macs on macOS 14 or later.
- Local hardware boundary: supported local modes are available on Apple Silicon Macs.
- Intel Macs: eligible Intel Macs can use cloud-backed models.
- Languages: fast local dictation for common languages, with broader language coverage where available.
- Pricing model: free trial, then monthly, yearly, or lifetime license. Subscriptions and lifetime cover all your devices.
If you want to compare this workflow against Apple’s built-in baseline, our guide on how to do speech to text on Mac walks through the setup and tradeoffs, and the Paraspeech pricing page shows how the monthly, yearly, and lifetime plans differ.
How Paraspeech works
- Hold your recording shortcut (Control by default).
- Speak naturally.
- Release to stop.
- Paraspeech transcribes using the mode you chose and inserts text where your cursor is
It also stores a local transcription history in the menu bar for quick reuse. If you’re working on confidential material, Privacy Mode clears history and stops saving new entries.
Developer workflows to test
Dictation is not “voice coding” (where you speak every symbol). It shines where developers already write in natural language:
- Commit messages & PR descriptions
Dictate a complete summary in 15–30 seconds, then edit for brevity. - Code review comments
Faster feedback loops: dictate rationale, edge cases, and suggested changes. - Tickets, bug reports, and repro steps
Dictate the narrative; attach logs/screenshots separately. - Architecture notes & ADRs
Talk through trade-offs while they’re fresh, then convert to structured bullets. - Documentation & onboarding
Especially for longform explanations, API docs, and runbooks.
Power-user setup (the part most people skip)
1) Choose the right insertion mode
- Use Auto-Send for chat, docs, tickets, and emails.
- Turn Auto-Send off for codebases where you want to review text before insertion. Clipboard-only reduces “oops, pasted into the wrong pane.”
If you want the exact install, permissions, and model-management flow before you start tweaking settings, the Paraspeech docs walk through the setup in more detail.
2) Turn on “Remove Fillers” Strip “um/uh”-type fillers automatically for cleaner drafts.
3) Build “Word Replacements” for your vocabulary Word replacements are the fastest path to “this understands my stack.” Start with:
- product names / acronyms:
api → API,kubernetes → Kubernetes - repos & codenames
- teammate names (proper casing)
- common technical terms that get misheard
Examples:
kubernetes -> Kubernetes
post hog -> PostHog
get hub -> GitHub
kuber netes -> Kubernetes
type script -> TypeScript
java script -> JavaScript4) Pick the right mode When processing location matters, record the active speech backend, the separate rewrite backend, and whether temporary cloud speech fallback was allowed or used. Eligible cloud-backed modes can provide broader coverage or support hardware without a local path.
5) Learn the “bail out” controls Recent releases add quality-of-life controls like:
- Double-tap ESC to cancel a recording
- Clear History button
- Saving recent recordings locally so you can re-transcribe if something fails
Privacy and processing boundaries
Paraspeech has local and cloud-backed processing paths. A selected mode alone does not prove where a session ran, so verify the active speech and rewrite backends and temporary speech-fallback state for the exact workflow. There are still a few points worth understanding:
- Permissions
- Microphone: required to capture audio; privacy policy notes the audio input is temporary during active transcription.
- Accessibility: required so Paraspeech can detect your hotkey and insert text; the privacy policy states no accessibility data is collected.
- Model download
- The app may connect to Hugging Face to download speech recognition models on launch; this may process technical request data (e.g., IP/logs). Models are stored locally.
- Cloud-backed modes
- Eligible cloud speech, including allowed temporary speech fallback, sends the audio needed for transcription to the selected service. Cloud rewrite sends the text needed for that operation.
- Analytics
- Analytics and diagnostics are separate operational flows; check the current privacy policy and settings for their exact scope.
- If opt-in is enabled, we mainly use PostHog to understand what we can improve.
If your main concern is preventing source code, credentials, or private client details from reaching a transcription or rewrite provider, require exact-run evidence of an active local speech backend, the intended on-device rewrite backend, no cloud speech fallback, and the relevant operational network state.
Where Paraspeech is not the best fit
- Older macOS (requires macOS 14 or later).
- Symbol-dense coding (regexes, heavy punctuation, lots of operator soup).
- Very noisy environments without a decent mic.
For those cases, use Paraspeech for the narrative parts of dev work (docs/tickets/reviews) and keep your normal keyboard flow for code-heavy edits.
2) Raycast - command palette + automation (macOS)
Raycast replaces a lot of “hunt for the thing” time with a fast launcher and an extension ecosystem. Strong for:
- app switching and search
- clipboard history
- snippets/text expansion
- scripts and workflow shortcuts
3) Keyboard Maestro - macros that eliminate repetitive friction (macOS)
When you catch yourself doing the same 5–10 click sequence multiple times per day, it’s macro time:
- window management and app orchestration
- text manipulation pipelines
- hotkeys that trigger shell scripts / AppleScript / UI automation
4) GitHub Copilot - code completion that’s good at boring parts
Copilot is most valuable when you treat it as:
- a boilerplate generator
- a “suggest the shape” helper
- a fast way to iterate on tests and edge-case handling
5) Cursor - AI-first editor for exploring and refactoring
Cursor can be useful when you want:
- repo-aware explanations (with guardrails)
- multi-file refactor assistance
- a workflow that blends chat + code edits without constant copy/paste
6) JetBrains IDEs - refactoring power tools
JetBrains IDEs are still hard to beat for:
- safe refactors across large codebases
- navigation and code intelligence
- debugging ergonomics for certain stacks (Java/Kotlin, Python, .NET, etc.)
7) VS Code - flexible “default” editor with deep extensions
VS Code earns its place via:
- a huge extension ecosystem
- remote dev / devcontainers workflows
- task runners + integrated debugging
8) Warp - modern terminal UX
Warp’s pitch is that the terminal can be more than a scrolling green screen:
- faster command re-use and navigation
- richer output handling
- optional AI features depending on how you configure it
9) OrbStack - fast containers on macOS
If Docker Desktop feels heavy, OrbStack is a popular alternative on macOS for:
- running containers with less overhead
- spinning up Linux environments locally
- improving the “containers are slow on Mac” experience
10) Postman - API testing without reinventing a client
When your job involves APIs, a dedicated client saves time on:
- repeatable request collections
- auth and environment management
- quick response inspection and debugging
11) Sentry - shorten the path from “bug report” to “root cause”
Observability is productivity. Sentry helps by:
- surfacing errors with stack traces + context
- connecting releases to regressions
- turning “can’t reproduce” into actionable data
12) Obsidian - notes that scale with your codebase
Obsidian is a strong choice when you want:
- local-first notes (Markdown)
- fast linking between ideas (ADR → ticket → PR)
- a personal knowledge base that doesn’t rot
A practical “stack” (if you want a starting point)
If you’re on macOS and want a minimal, high-ROI setup:
- Paraspeech for writing-heavy dev work (reviews, tickets, docs)
- Raycast for command palette + quick actions
- Keyboard Maestro for repeatable automation
- VS Code or JetBrains for your main editor
- OrbStack if containers are part of your daily flow
- Sentry if you ship software used by humans
Sources (Paraspeech deep dive)
- Paraspeech homepage: https://paraspeech.com/
- Paraspeech documentation: https://paraspeech.com/docs
- Paraspeech changelog: https://paraspeech.com/changelog
- Paraspeech privacy policy: https://www.iubenda.com/privacy-policy/72579598
- Paraspeech terms: https://www.iubenda.com/terms-and-conditions/72579598
