Over the past few weeks, much to my surprise, I have discovered the pleasure and power of voice dictation. I never expected to adopt this way of writing, let alone use it often enough to devote an article to it. Yet it has gradually found its way into my daily life: replying to a message, giving an AI tool an instruction, or quickly capturing an idea without immediately returning to the keyboard.
I do not see dictation as a replacement for writing. For a long, precise, and structured piece of text, I still prefer to write at the keyboard. But much of what we write every day does not require that level of polish. In those situations, speaking for a few seconds and then reviewing the result is often smoother than typing everything out.
A shortcut for writing in any application
The principle is very simple. A tool is installed on the computer and assigned to a keyboard shortcut. I press that shortcut, dictate my text, and the application transcribes it. Depending on the tool and its configuration, the result is either inserted directly into the active field or placed on the clipboard so that I can paste it wherever I want.
That last step matters: dictation is not confined to one particular application. I can use the text in an email, a chat, a document, an AI tool, or even in code when I need to write an instruction or a comment. The result remains text, with all the practical benefits that entails: I can review it, correct it, move it, or save it.
I started with Wispr Flow, which I found very convincing and whose free plan already lets you test the idea quite generously. I then switched to Open Whisper. Its main advantage is that it can run for free, locally, and without an account. This setup is more comfortable with a sufficiently powerful GPU, however.
There is also a hybrid approach: the interface remains Open Whisper, but the transcription is handled by an external service using an API key, for example from OpenAI or Gemini. The tool itself may then be free, while API usage is billed under the chosen provider’s terms. The right compromise therefore depends on your computer, how often you use it, and how important local processing is to you.
Voice dictation that remains asynchronous
To begin with, I am really not a fan of voice messages. The ones I receive on WhatsApp particularly annoy me: the sender may save some time, but they ask the recipient to listen to a one-, two-, or three-minute recording without being able to scan its contents. It is harder to find a specific piece of information, listen discreetly, or gauge at a glance how much effort a reply will take.
This constraint resembles the one imposed by calls and voicemail, which I have already discussed in my article about why I no longer answer the phone. In both cases, the person receiving the message has to adapt to the pace chosen by the sender.
The kind of dictation I am discussing here is different. Voice is only the input method; the recipient ultimately receives a written message. Communication therefore remains asynchronous, and the reader retains all the benefits of text. They can skim it, return to a sentence, copy a particular detail, search for a word, or reply when it suits them.
In other words, I gain the ease of speaking without passing the burden of listening to a recording on to the other person. That is probably the main reason this use has won me over while voice messages still irritate me.
As much about posture as speed
Voice dictation is often presented as a way to write faster. That is sometimes true, but it is not its only benefit. For me, it mainly addresses the very practical matter of posture.
A good developer will tell you to master shortcuts, keep your hands on the keyboard, and type efficiently with all ten fingers. In theory, I agree. In practice, we also spend a lot of time using a mouse, moving between interfaces, or looking at something while sitting in a position that is not ideal for typing.
At those moments, returning to the keyboard just to enter a simple sentence creates a small interruption. It is nothing dramatic, but it happens dozens of times. With dictation, I can stay where I am, trigger the shortcut, and say exactly what I want to write. The action requires less preparation and lets me maintain the flow of whatever I am doing.
This freedom is also pleasant when I want to step back from the screen or give my hands a brief rest. I am not trying to banish the keyboard; I simply appreciate having another input method when it is better suited to the situation.
Start with simple text
I find voice dictation particularly powerful for relatively basic content: a short reply, an instruction, a question, an idea to capture, or a few sentences whose structure is already clear in my mind. In these cases, the benefit is immediate. I speak, quickly check the transcription, and send or paste it wherever I need it.
At first, I thought that limit would remain fairly strict. I could not imagine dictating more than a few lines without losing my train of thought or producing a confusing text. Yet I now find myself creating increasingly long and complex passages.
The key is to take your time. There is no need to speak without stopping as though you were recording a radio show. I can pause, think for a few seconds, resume, and rephrase. These pauses let me structure my thoughts before continuing. The first draft is not always elegant, but it is often organized enough to provide a solid foundation.
Dictation also exposes imprecise thinking rather quickly. When a sentence becomes endless or I repeat the same idea three times, the problem is not always the transcription: sometimes I simply did not yet know exactly what I wanted to say. Reviewing the result therefore remains essential, especially when the text is intended for someone else.
Do not expect it to replace writing
For content that requires precise construction, carefully controlled rhythm, or numerous references, I still prefer to write. A keyboard makes it easier to go back, move paragraphs around, and make the small edits that help an argument develop. It also lets me see the structure taking shape as I write.
Dictation is more likely to produce an oral first draft, with repetitions, weak transitions, and sentences that sometimes run too long. Even when the transcription is excellent, it does not automatically turn spontaneous speech into publishable prose. You still have to review, cut, clarify, and sometimes rewrite an entire passage.
But probably 90% of what we write in daily life is not meant to be great literature. A message to a colleague, a practical request, or an instruction to an AI mainly needs to be understandable. For these uses, an accurate transcription followed by a quick check suits me perfectly well.
This article is a good illustration: I made extensive use of voice dictation to produce its raw material. That did not spare me from structuring the argument and working on the prose, but it did allow me to put down some ideas more naturally before revising them.
Built-in features and system-level tools
Many AI tools, including ChatGPT and Gemini, already offer a dictation feature directly in their interface. Everything I have tested so far does the job properly. If your only need is to speak to an AI instead of typing your prompts, that built-in feature may be more than enough.
A tool such as Wispr Flow or Open Whisper nevertheless offers one important advantage: it operates at the level of the operating system. The same shortcut remains available regardless of which application is open. I do not have to send my text through a chat and then copy it to its final destination; I can dictate it directly wherever I am working.
Depending on the product, these tools also offer a dictation history and personalized recognition of certain words. The history can be reassuring if a transcription disappears or if you want to retrieve something you dictated recently. Personalization is useful for proper names, professional vocabulary, or terms that the engine struggles to understand. I am not sure these features are decisive for everyone, but they do contribute to the overall convenience.
This centralization also deserves some attention. A dictation history can contain sensitive personal or professional messages. Before adopting a tool, it is therefore sensible to check where recordings and transcriptions are processed, what is retained, and how to delete the history. Open Whisper’s ability to run locally may be particularly appealing to people who want greater control over their data.
What about mobile?
Equivalent solutions are available on iOS and Android, alongside the dictation features already built into mobile keyboards. On paper, a phone seems like a natural environment for this use: we often have one in our hand without being in a good position to type a longer piece of text.
I have not yet taken the time to test these versions seriously, however. I would therefore rather not offer a definitive opinion on their quality, their integration, or their value compared with native features. For now, my experience mostly concerns computers, where the global shortcut and the ability to insert text into any application genuinely change my workflow.
A modest tool that has become second nature
Voice dictation has not transformed the way I write, and it replaces neither the keyboard nor the editing process. It has simply removed a series of small points of friction. When a sentence is clear in my mind, I can now put it down immediately, even if my hands are not already on the keyboard.
I recommend starting in the same way: a few short messages, simple instructions, or ideas to capture, without immediately trying to dictate an entire article. You gradually learn to speak in a more structured way, leave pauses, and recognize the situations in which the tool genuinely helps.
Perhaps the most convincing sign is that I barely think about the technology itself anymore. I press a shortcut, speak, and the text appears. For a productivity tool, that unobtrusiveness is often the best evidence that it has found its place.