Dictate on your Mac.
Not on someone's server.
my_voice_ is your own private dictation. Hold a key, speak, and your words paste into whatever you are typing in. The audio never leaves your machine. A small monthly, a cheaper year, or own it once.
The tool you talk to all day should not be renting you your own voice.
If you are a creator or an indie coder, you dictate constantly: commit messages, docs, prompts, replies. Every word is your work. The popular cloud dictation tools take that stream, send it to their servers, charge you every month, and cap you until you pay. my_voice_ does the opposite by construction.
It stays on your machine
Whisper runs locally. Your audio is transcribed on your own device and never leaves it. No server, no telemetry, nothing to harvest.
Own it, don't rent it
A small monthly, a cheaper year, or pay $107 once and keep it. No forever-subscription for something that should simply run on your computer. No word cap.
No account, no login wall
Install it and use it. There is no sign-up, no cloud identity, nothing between you and your own words.
It learns, privately
Your names, brands and corrections get sharper the more you use it. That learning lives on your machine, not in a training set.
Privacy is not the trade-off. It is the unlock.
Because my_voice_ never phones home, it does things a cloud tool structurally cannot. Their business needs the server, the account and the meter. Ours needs none of it, so you get more, not less.
No caps, ever
No per-word server bill means no 2,000-word weekly wall and no usage meter. Dictate all day, every day, and it costs the same: nothing more.
Works with no internet
Plane, cafe dead zone, locked-down network. Whisper runs on your machine, so it keeps working when a cloud tool simply stops.
It learns you, and only you
A personal dictionary of your names, brands and corrections that sharpens with use and never enters a shared training set. A cloud tool cannot offer that without harvesting you.
Say anything
Medical notes, legal matters, a password read aloud, a raw first draft. It never leaves the machine, so you never self-censor for the server.
It is not just your words. It is you.
Dictate into a cloud tool all day and you are not handing over "some text." You are handing over three things at once, and most people never clock the last two.
Your ideas
Everything you look up, draft and think out loud. Every prompt, plan and half-formed thought, the raw material of your work, streamed to someone else's servers.
Your identity
Your voiceprint is a biometric, and with a few seconds of audio, voice cloning is now consumer-grade. You are handing over a key that can imitate you.
Your telemetry
Voice carries stress, mood and fatigue. Emotion-from-voice is a real industry and voice health signals are an active research field. Day after day, your voice reads how you are doing.
And it catches more than you think. A cloud recorder does not wait for you to hit send: the words you dictate and then delete, the thing you said in anger and thought better of, were already transcribed off your machine. It picks up the background too, other voices, the kids in the room. Text tells them what you did. Your voice tells them who you are and how you are doing. When you hold that key down, you might as well be holding the net open for them.
And it is not hypothetical.
Anything that leaves your machine is held by someone else, which means it can be subpoenaed, breached, or used in ways you never agreed to. This is on the record, not a scare.
Cloud transcription, in court
Otter.ai, a cloud transcription tool, faces a class-action filed in August 2025 alleging it "deceptively and surreptitiously" records private conversations and uses them to train its AI without consent. Unproven allegations, but the exposure is the point.
Your data, subpoenaed
Cloud voice data has been sought as evidence before, from Amazon Echo recordings onward. If a provider holds it, a court can ask for it and a breach can leak it.
my_voice_ has nothing to hand over
The audio never leaves your machine, so there is no server copy to subpoena, breach or sell. This is information hygiene by construction: privacy is the architecture, not a promise.
Otter.ai class-action: filed U.S. District Court, N.D. California, August 2025 (npr.org, gpb.org, read 2026-07-20), unproven allegations. Amazon Echo evidence: State of Arkansas v. Bates (pbs.org, cnn.com). Corrections welcome: admin@lionsedge.tech
my_voice_ vs Wispr Flow, on the facts
Wispr Flow is a fine cloud tool. The difference is architecture and how you pay, and it is checkable from Wispr Flow's own public pages.
| my_voice_ (Local Gorilla) | Wispr Flow | |
|---|---|---|
| Where transcription runs | On your machine (local Whisper) | In the cloud |
| Your audio | Never leaves your machine | Sent to its servers to transcribe |
| Works offline | Yes | No, needs a connection |
| Pricing | $5.30/mo, $53/yr, or $107 once | Subscription only: US$15/mo, or US$12/mo annual |
| Own it outright | Yes, $107 once | No |
| Free tier | n/a, own it | Capped at 2,000 words per week |
| Account required | No | Yes |
| Platforms | macOS now, Windows coming | Mac, Windows, iPhone, Android |
Wispr Flow facts from wisprflow.ai/privacy and wisprflow.ai/pricing, read 2026-07-20. Wispr Flow runs on more platforms today; Windows for my_voice_ is on the roadmap. Corrections welcome: admin@lionsedge.tech
One price. Every platform. All under the competition.
A small monthly, a cheaper year, or own it once and keep it forever. No per-word meter, no Mac tax, no Windows tax. Prices do not move.
Every plan starts with a 7-day free trial. No card to look around, no auto-charge sprung on you. Then pick monthly, yearly, or own it.
Wispr Flow Pro is US$12 to US$15 a month, US$144 to US$180 a year, every year. Own my_voice_ outright for $107, less than a single year of Wispr, and it never charges you again.
Comparison: Wispr Flow Pro US$12-15/mo (wisprflow.ai/pricing, read 2026-07-20). All prices USD, one price on every platform.
Questions
Is there a local, private alternative to Wispr Flow?
Yes. my_voice_ runs Whisper on your machine, so your audio never leaves it. Wispr Flow's own privacy page states transcription always happens in the cloud (read 2026-07-20). my_voice_ has no account, and you can own it outright.
Does Wispr Flow work offline or keep my audio on my device?
No. Per its privacy page (wisprflow.ai/privacy, read 2026-07-20), Wispr Flow transcription always happens in the cloud, so it needs a connection and your audio is sent to its servers. my_voice_ transcribes on your machine and works offline.
How much does my_voice_ cost?
Three options, all under the competition: US$5.30 per month, US$53 per year, or own it outright for US$107. Wispr Flow is subscription-only at US$15 per month (or US$12 per month billed annually). One price on every platform.
What is the best on-device dictation app for Mac?
my_voice_ runs Whisper entirely on your machine. You hold a key, speak, and the text pastes into whatever app you are in. No cloud, no account, no telemetry, and you can own it once instead of a monthly subscription.
Does my voice leave my computer with my_voice_?
No. The audio is captured and transcribed on your own machine and never leaves it. There is no server, no account and no telemetry. That is the architecture, not a setting you toggle.
Is my_voice_ available for Windows?
macOS (Apple Silicon) ships first; Windows is on the roadmap. The goal is system-agnostic, so the same private dictation runs wherever you work. Local Gorilla's browser extensions are already cross-platform.
What can a cloud dictation app actually do with my voice?
Once your audio is on a provider's servers, it is three things at once: your ideas (the content), your identity (a voiceprint is a biometric, and voice cloning now needs only seconds of audio), and telemetry (voice carries stress, mood and health signals that a longitudinal recording can read). Cloud transcription also captures what you dictate then delete, and background voices. my_voice_ transcribes on your machine, so none of it leaves.
Has cloud transcription data ever been used against people?
Cloud-held data is subpoenable and breachable by nature. Otter.ai, a cloud transcription tool, faces a class-action filed in August 2025 alleging it recorded private conversations and used them to train its AI without consent (unproven allegations). Cloud voice data such as Amazon Echo recordings has also been sought as court evidence. Because my_voice_ keeps audio on your machine, there is no server copy to hand over.
Does a local dictation tool have fewer features than a cloud one?
The opposite. Because my_voice_ has no server to pay for, it has no per-word meter or weekly word cap, it works fully offline, and it learns your personal vocabulary without sending it to a shared training set. And you can dictate anything sensitive because it never leaves your machine. Those are features a cloud tool structurally cannot match.
Own your voice.
Launching soon, macOS first. Leave your email and you will be the first to know.
Notify me at launch