← All articles

Blog / Dictation

Choosing a voice input method on a Mac

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.

Decision guide

Mac voice input methods by job
OptionChoose it whenCheck first
Apple DictationYou want spoken text placed at the insertion point in an app you are already using, with no new software.Whether Dictation on your Mac is processed on-device or sent to Apple, shown in Keyboard settings.
Voice ControlThe job is operating the Mac by voice: navigating, clicking, and editing with spoken commands.That you are prepared to learn command vocabulary; this is broader than entering one paragraph.
Separate voice input layer (Relay for macOS)You speak in natural, revised, incomplete thoughts and want text or actions inside the app where you work.The product’s data-processing boundary and system requirements; Relay needs an M-series Mac.
Dedicated transcription app (MacWhisper)The output is a transcript and you want a transcription-oriented workflow rather than the OS feature.How the app handles audio and where transcription runs, per its documentation.
Conversational assistant (ChatGPT Voice)The conversation is the work and the words do not need to land at a cursor first.That the transcript reaches the text chat after you exit; moving it into another app is a separate step.

This is a documentation-based guide, checked against Apple’s Dictation, Voice Control, and sound input guides, MacWhisper’s dictation documentation, OpenAI’s Voice FAQ, and Relay’s product and privacy information on September 4, 2026. Raymond Zhu is a founding member of Relay. Sources: Apple’s Dictation guide, Apple’s Voice Control guide, Apple’s sound-input settings guide, MacWhisper Dictation, ChatGPT Voice, Relay for macOS, Relay Privacy Policy.

Start from the job, not the tool.

Most disappointment with voice input on a Mac comes from asking one method to do a job it was not designed for. Speaking a sentence at a cursor, controlling the desktop, thinking out loud until an idea is usable, and producing a clean transcript are four different tasks. Each of the methods below is good at one or two of them.

So start by naming the job you do most. If the answer is “put these words here,” the built-in option is probably enough. If the answer is “get this half-formed intention into the app I am using without a second editing pass,” you are describing a separate voice input layer. If the answer is “run my Mac without a keyboard,” that is Voice Control.

Sources: Apple’s Dictation guide, Apple’s Voice Control guide.

Direct speech-to-text at the cursor.

Apple Dictation is the built-in way to speak text wherever your Mac accepts typing. Apple documents starting it with the Microphone key, a Dictation shortcut, or Edit > Start Dictation, then speaking at the insertion point in the app you are already using. It is the smallest change to an existing workflow, which is why it should be the first thing you try.

Its privacy behaviour is a setting rather than a slogan: Apple says the Keyboard settings show whether general text Dictation is processed on-device or sent to Siri servers, and that the optional Improve Siri & Dictation setting controls whether recordings may be stored and reviewed. Check those on the Mac you use.

When built-in Dictation is not enough, the two most common reasons are that you want a different kind of input, covered in the next sections, or that you prefer a dedicated app’s workflow. The detailed comparisons are Apple Dictation vs Relay for macOS and MacWhisper vs Apple Dictation on Mac.

Sources: Apple’s Dictation guide, Apple’s Siri, Dictation & Privacy information.

Voice control for operating the interface.

Voice Control is a separate macOS feature for using spoken commands to interact with the desktop and apps. Apple describes navigating, clicking items on screen, dictating, and editing text by voice. It is the right method when the job is running the Mac rather than entering a paragraph, and it is often the best fit for people who cannot rely on a keyboard or pointer.

Because Dictation and Voice Control are both built in, many people end up using both: Voice Control to move around, Dictation or another input method for the text itself. Apple Dictation vs Voice Control on Mac covers where each one stops.

Sources: Apple’s Voice Control guide, Apple’s Dictation guide.

A separate voice input layer inside the apps you already use.

A third category sits between transcription and a chatbot: a voice input layer that stays in the app where your cursor is but accepts natural speech, revisions, and incomplete thoughts, then turns them into text or actions there. Relay for macOS is built for this. It is not a separate writing destination and not a general-purpose assistant; it behaves more like a keyboard that understands what you meant.

Choose this category when the input is more than a clean transcript: shaping a message for the person or app it is going to, turning a spoken instruction into a Terminal command, or triggering a workflow. The trade-off is adopting another product and reading its data-processing boundary. Relay captures audio only while its recording key is held, keeps context optional and local until you choose to share it, and states that it does not retain processing data; the Privacy Policy is the authoritative source.

Two clarifications come up often. Relay for macOS is standalone software for M-series Macs; the Relay Q hardware is optional, as Does Relay require Relay Q? explains. And a conversational assistant is a different job again: ChatGPT Voice is designed for a spoken conversation whose transcript is added to the text chat after you exit, which ChatGPT Voice vs Mac Dictation vs Relay compares. For the basics, start with What is voice input for macOS? and How Relay handles context and privacy.

Sources: Relay for macOS, Relay Privacy Policy, ChatGPT Voice.

Dedicated transcription workflows.

If the deliverable is a transcript, a transcription-oriented app can fit better than the operating system feature. MacWhisper documents a Dictation feature that transcribes into a text field directly, alongside its file-based transcription workflow. The reason to choose it is the workflow and the app’s own documented handling of audio, not a claim that it is more accurate; this guide makes no accuracy comparisons.

The practical test is where the words need to end up. Text that belongs at a cursor in an app you already have open argues for Dictation or a voice input layer; a recording that needs to become a document argues for a transcription app. MacWhisper vs Apple Dictation on Mac works through that boundary.

Sources: MacWhisper Dictation, Apple’s Dictation guide.

Microphone and activation choices.

Whichever method you pick, two setup decisions sit underneath it. The first is audio: macOS lets you select an input device and view its input-level meter in Sound settings, so start with the microphone your Mac already has and add a headset or external microphone only for a specific reason such as shared space or repeatable placement. External microphone vs headset vs Mac microphone for dictation and The best computer accessories for voice input cover this without a shopping list.

The second is activation. A keyboard shortcut is fastest when your hands are already on the keys; a push-to-talk control makes the start and end of every input explicit, which matters in shared spaces and for people who do not want a system waiting for a wake phrase; voice activation trades that explicitness for convenience. Push-to-talk button vs keyboard shortcut vs voice activation compares the three.

Sources: Apple’s sound-input settings guide, Apple’s Dictation guide.

Keyboard, accessibility, and voice as complementary inputs.

Voice input rarely replaces the keyboard outright. Passwords, code, spreadsheets, and precise edits usually stay on the keys, while messages, notes, and first drafts are where speaking pays off. If typing feels slow, the cause matters more than the accessory: I’m a slow typer: what can I do? separates those causes, and Mechanical keyboard vs ergonomic keyboard vs voice input treats voice as one option among three rather than a replacement.

Capture away from the Mac is a separate question. A phone is usually the right tool for a quick thought, and AI glasses or a dedicated device only earn their place for specific situations; AI glasses vs a phone vs a dedicated AI device and Ray-Ban Meta glasses vs a phone for capturing ideas cover that decision. Once the idea reaches your desk, this guide takes over.

Sources: Apple’s Dictation guide.

How this guide was made.

This is a documentation-based guide, not a benchmark. It describes each method’s stated purpose, activation, and privacy boundary as documented by its maker on September 4, 2026, and links to the detailed comparison for each decision. It makes no claims about accuracy, speed, or language coverage.

Relay has a commercial interest in one of the categories described here: Raymond Zhu, the author, is a founding member of Relay. That is why the built-in options come first, and why the guide recommends trying them before adopting any separate product, including Relay.

Sources: Apple’s Dictation guide, Relay for macOS.