Blog
Relay is a voice interface made by Relay Voice Ltd, a company registered in England and Wales. Relay for macOS is available now and Relay for Windows is in public Beta; Relay Q is optional push-to-talk hardware that ships in January 2027. Relay Skills are voice-triggered actions, and Relay Unlimited is the paid plan.
Relay Q is a modular push-to-talk voice input device from Relay Voice Ltd. You press and hold the top to speak, release to get the result, and double-press to send it, in the apps you already use. It’s designed to work as a handheld controller or a magnetic clip-on microphone, is optional for the Relay apps, and ships in January 2027.
Relay for macOS starts with 14 days of Relay Unlimited, with no credit card required. Relay Unlimited then costs US$15 a month; if you don’t subscribe, your account moves to the Free tier. The other paid option is the US$150 Early supporter bundle, a one-time purchase with 12 months of Relay Unlimited and Relay Q hardware that ships in January 2027.
Choose Wispr Flow if you want polished dictation on Mac, Windows, iPhone and Android with per-app Styles and Command Mode. Choose Relay if you work on a Mac and want speech shaped around your final intent, Terminal commands, inserted links or custom Relay Skills. Both are cloud services, and both cost US$15 a month on monthly billing.
Dictation is built into every Mac. Turn it on in System Settings > Keyboard > Dictation, place the cursor where you want text, and start it with the Microphone key, your Dictation shortcut or Edit > Start Dictation. You can say punctuation and formatting commands, switch languages and choose which microphone it uses.
Context-aware dictation uses information beyond your voice, such as the app you are typing in, selected text, a window screenshot or your saved names and terms, to shape what gets written. It helps most when the same words need a different form in different places. Several tools offer it, so compare what each one reads and when.
Speaking a prompt makes it easier to give an AI the background it needs. ChatGPT and Claude have built-in dictation that turns speech into a text prompt you can review, separate from their voice conversation modes. A system-wide voice layer helps when you prompt across several apps. Structure what you say as goal, context, constraints and format.
Most AI coding tools now have voice input of their own: Claude Code has /voice, Cursor has a microphone in its chat input, and Microsoft’s VS Code Speech extension adds local voice to Copilot Chat. A system-wide voice layer matters when you move between the editor, the Terminal, the browser and chat. Review any dictated command before it runs.
For dictation that works without the internet, start with Apple Dictation’s on-device mode for supported languages, or a local-model app such as Superwhisper, MacWhisper, VoiceInk, Handy or Spokenly. Check both stages, speech recognition and any AI rewrite, because a local speech model does not make a cloud rewrite local. Relay processes requests in the cloud, so it isn’t an offline option.
For short messages, yes, in at least one well-designed study: speech entry on a phone was about three times faster than the phone keyboard, with fewer errors made along the way, though it left slightly more errors in the final text. That result doesn’t transfer automatically to a desktop keyboard, long documents or editing time, so the honest answer for your own work comes from timing a real task.
Start with Apple Dictation for simple built-in text entry. Shortlist Wispr Flow for writing across devices, Aqua Voice for technical dictation, Superwhisper for model control, MacWhisper for transcripts, or Relay for contextual text and Skills. The best choice depends on the work you need to finish.
Mac voice input is not one feature. Dictation, Voice Control, a separate voice layer, and transcription apps solve different jobs. This guide helps you pick by job, then routes you to the detailed comparisons.
Use the phone you already have for most captured thoughts. AI glasses make more sense when the task depends on what you can see while moving; a dedicated device only earns its place when physical activation is the real problem.
Use a phone when you need a private, reviewable note. Ray-Ban Meta glasses are more useful when you are moving and the idea is tied to what you can see, hear, or do hands-free.
First identify whether the bottleneck is keyboard familiarity, physical comfort, accuracy, or getting a developing thought into a computer. Practice, accessibility settings, dictation, and a different keyboard each solve different problems.
Choose a mechanical keyboard for the typing feel and layout you prefer, an ergonomic setup when physical comfort is the central issue, and voice input when getting a developing thought into a draft is the real bottleneck. Most writers benefit from a mix.
Start with the Mac microphone if it is comfortable and clear enough in your normal workspace. Choose a headset or external microphone for placement, privacy, and routine, not because any accessory guarantees better dictation.
Use a keyboard shortcut at a desk, a push-to-talk control when you need a deliberate start and stop, and voice activation only when its convenience outweighs its social and privacy trade-offs.
Start with the Mac, microphone, and shortcut you already have. Add a headset, external microphone, or physical control only after you can name the exact moment your current setup makes voice input harder.
Use Apple Dictation when you want to put spoken text at the cursor. Use Voice Control when you need to navigate, click, edit, and control the Mac with speech, not just dictate a sentence.
Use Apple Dictation for the built-in path to speak text at the cursor. Use MacWhisper when its dedicated transcription and dictation workflow, plus its local or cloud processing choices, better matches the work you need to do.
Use Mac Dictation to put text at the cursor, ChatGPT Voice to have a spoken conversation with ChatGPT, and Relay for macOS when you want a separate voice-input layer inside the Mac apps where you work.
Apple Dictation is the built-in choice for straightforward text entry. Relay is a separate M-series Mac interface for context-aware voice input; choose based on workflow and privacy settings.
Voice input for macOS lets you speak text and actions into the app where you work. Relay is built for that workflow on M-series Macs.
No. Relay for macOS is standalone software; Relay Q is an optional hardware companion with its own release timeline.
Relay records only after you start capture with its recording key, in hold or toggle mode, and it uses screen and app context only for the request you make. Audio and any context it needs pass through Relay to AI providers for processing and are not stored on Relay’s servers. It is not an offline app.