What do you like best about VoiceX?
The core loop is what won me over: cursor in any app, one press of the Fn key, speak, and clean text appears where the cursor sits. No separate window, no transcript to copy out of, no app-by-app setup. The first afternoon I used it I dictated into Mail, Slack, a Notion doc, a browser form, and a ChatGPT prompt box, and the behavior was identical in all five. That consistency is the whole product in one sentence. A dictation tool that only works in its own editor is a note-taking app with extra steps. VoiceX works wherever I already am.
The grammar polishing is the part that separates it from the dictation built into the operating system. Raw speech is messy. I ramble, I restart sentences, I say "uh" and "so basically" and correct myself mid-thought. Apple's built-in dictation transcribes all of that faithfully, which means every dictated paragraph needs an editing pass before anyone can see it. VoiceX takes the same rambling input and returns prose I would actually send: filler stripped, sentences closed properly, punctuation where it belongs. The output is not a transcript of what I said. It is what I meant to write. For email in particular that difference decides whether dictation is usable at all, because nobody wants to send a client a message that reads like a court stenographer's record of them thinking out loud.
Language auto-detection matters more to me than it will to a monolingual user, and it is handled better here than in anything else I have tried. I work in Italian and English all day. Internal messages, calls with local clients, notes to myself are Italian. Reviews, outreach, and most product documentation are English. VoiceX detects which language I am speaking and writes in that language, with no menu, no toggle, no "switch input source" ritual. I have dictated an Italian Slack message and an English email back to back without touching a setting. The claimed coverage is 99 languages. I can only vouch for two, but for those two the switching has been invisible, and invisible is exactly what a bilingual workflow needs.
The personal dictionary earns its keep quietly. Product names, client names, and the SaaS vocabulary I use constantly used to come out mangled: tool names turned into dictionary words, brand spellings normalized into something wrong. I added my recurring terms once and the mangling mostly stopped. It is not a glamorous feature. It is the feature that decides whether dictated text needs proofreading for proper nouns, and after a week of seeding the dictionary, mine largely does not.
Speed is the obvious pitch and it holds up in practice. I speak a first draft several times faster than I type it, and because the polishing removes the cleanup pass, the gap does not get eaten back on the editing side. A three-paragraph email that used to take a few minutes of typing now takes the time it takes to say it plus a quick read-through. Over a day of heavy written output that compounds into something noticeable, though I will not pretend to have measured it.
The privacy posture is stated plainly and that matters for the work I dictate. Voice recordings are not stored, and the company is explicit that voice data and transcriptions are not used to train models. A meaningful share of what I speak into this thing is client business: pipeline notes, partnership terms, draft proposals. Dictation tools that are vague about what happens to audio are disqualified for that material by default. VoiceX answers the question directly on its own pages, and the Enterprise tier carries the compliance items (SOC 2 Type II, ISO 27001, DPA) that a larger organization would need before rolling this out. I am not on that tier, but its existence tells me the company has thought past the solo user.
A few smaller things that make the daily experience better than the feature list suggests:
- Onboarding is close to zero. Install, grant microphone permission, press Fn, talk. I had a colleague producing usable text inside five minutes with no explanation from me.
- The trigger is a single physical key, not a chord. When my hands are half-occupied, one keypress is the difference between using dictation and not bothering.
- Output lands at the cursor as ordinary text. Whatever app I am in treats it as if I typed it, so formatting, mentions, and send shortcuts all behave normally afterward.
Pricing sits at a level where the decision is easy. Pro is 10 dollars per user per month for unlimited words, and the free tier is a real free tier with 2,100 words a week rather than a seven-day trial in disguise. For a tool that touches nearly every piece of text I produce, that price is not a factor I spend time thinking about. The yearly option shaves two months off, which is what I ended up on. Review collected by and hosted on G2.com.
What do you dislike about VoiceX?
Platform coverage is the honest first answer. This is a Mac product today. Windows is listed as coming soon, and until it ships, half of any mixed-platform team simply cannot use it, which blocks the shared dictionary and the team side of the Pro plan from being useful for us. The mobile apps exist for iOS and Android, but the experience there is younger than the desktop one and the pricing page itself still marks mobile as in progress, so I treat the phone apps as a bonus rather than a place I rely on. My actual workflow is fine, because I live on a Mac. My recommendation to colleagues comes with an asterisk until the Windows build lands.
The polishing occasionally polishes too hard. Most of the time the cleanup is exactly right, but when I dictate something deliberately informal, or a sentence containing an exact string I need preserved (a feature name phrased a specific way, a fragment I want left as a fragment), the AI sometimes normalizes it into tidier prose than I asked for. The fix is quick, I edit the line or re-dictate it more explicitly, and the personal dictionary prevents the proper-noun cases. But it is a real behavior of the tool: the same intelligence that makes messy speech sendable will sometimes flatten intentional roughness. I would take a per-dictation switch between "polish" and "verbatim" over almost any other feature request I have.
The free tier cap is tight for a genuine evaluation. 2,100 words a week sounds like a lot until you realize a normal working day of email and messages burns through it before Wednesday. I hit the ceiling in my first week, which pushed me onto Pro faster than I had planned. That is presumably the point of the cap, and Pro is cheap enough that I am not aggrieved, but someone trying to evaluate the tool against a full week of real usage cannot do it on the free plan. A trial period of unlimited use would give a truer picture than a word budget does.
Some capability lives behind Pro that I would argue belongs lower. Command mode and the enhanced transcription tier are Pro-only, and enhanced transcription in particular is close to the core promise of the product. The free tier is genuinely usable for light dictation, so this is a quibble about where one line sits rather than a complaint about a crippled free plan, but it is the line I would move.
One setup note rather than a flaw: the Fn key already has default behaviors on macOS, so first-run involves a brief negotiation with system settings to make sure a single press goes to VoiceX and not to the emoji picker. Took two minutes and has not come up since. Worth knowing before you assume something is broken. Review collected by and hosted on G2.com.
