Introduction
If you've been shopping for voice-to-text tools lately, you've probably noticed the same pattern I have: everything is a subscription, everything runs in the cloud, and everything wants your audio on someone else's servers. Hitoku Draft takes the opposite bet — a $5 one-time purchase that runs entirely on your Mac, with zero data leaving the device.
I've spent time with it, and this review is the builder-to-builder take: what it actually does, where it breaks down, and whether it's worth the impulse-buy price tag. Short version — it's a genuinely interesting tool with real trade-offs. Let's get into them.
Key Features
Hitoku Draft is narrow by design. It's not trying to be Descript or a full transcription suite. It's a focused local voice-to-text and AI editing utility. Here's what you're actually getting:
- On-device speech-to-text — no cloud calls, no API keys, no round-trip latency to a server. Everything runs on your Mac.
- Hotkey activation anywhere — press a shortcut in any Mac app with a text field and start dictating. Slack, Notes, VS Code, browser inputs — it works system-wide.
- Local AI text editing — cleanup, rewriting, and light editing run through the Apple Neural Engine and GPU in parallel, so you're not just getting raw transcription.
- Audio and video file transcription — drop in a file and get text out, still without touching the internet.
- Multi-language support — English, French, German, Spanish, Japanese, Arabic, Chinese, and more, all handled on-device.
- Zero telemetry, no account — you download it, you use it, that's it. No signup, no dashboard, no tracking.
The technical bet here is straightforward: Apple Silicon is fast enough now that you don't need a cloud GPU to do decent transcription and light LLM work. In practice, that bet mostly pays off — inference feels snappy on an M-series chip.
Pricing Breakdown
This is the easy part. There's one plan.
| Plan | Price | What You Get |
|---|---|---|
| Launch | $5 one-time | Full app, all features, all languages, lifetime use |
No subscription, no tiers, no upsells. At $5 it's cheaper than a single month of most cloud-based voice tools. Compare that to Wispr Flow's monthly subscription or Deepgram's usage-based API pricing and the math is obviously favorable — if the feature set actually fits your workflow.
Worth flagging: this is a launch price. It may not stay at $5 forever. There's also no free trial, so you're paying blind. At $5 that's a low-risk bet, but it's still worth knowing before you click buy.
Pros & Cons
Pros
- Complete privacy — if you work with sensitive material (client calls, medical notes, legal drafts, unreleased product specs), nothing leaves your machine. This is the real selling point.
- One-time purchase — no subscription fatigue, no auto-renewal to forget about.
- Works everywhere on Mac — the system-wide hotkey means you're not locked into a specific editor or web app.
- Fast local inference — Apple Neural Engine + GPU in parallel means transcription doesn't feel sluggish.
- Fully offline — works on a plane, in a SCIF, on a spotty hotel wifi. Doesn't matter.
Cons
- Apple Silicon Mac only — if you're on Intel Mac, Windows, or Linux, this isn't for you. Full stop.
- Single platform — no iOS, no cross-device sync, no web version. Mac or nothing.
- Early-stage product — limited track record, small user base, and you're partly betting on the developer sticking around to ship updates.
- No free trial — $5 is low risk, but you can't kick the tires before paying.
- Narrow scope — this isn't a full transcription platform with speaker diarization, timestamps export, team collaboration, etc. It's a writing tool.
Who Is It For
Hitoku Draft has a specific user in mind, and being honest about that matters more than pretending it's for everyone.
Buy it if you are:
- A Mac user on Apple Silicon (M1 or newer) who dictates a lot of text — emails, docs, code comments, Slack messages.
- Privacy-focused by default — journalists, lawyers, therapists, founders working on unreleased products, anyone under NDA.
- Tired of subscriptions and want to own your tools.
- Someone who works offline often or in restricted network environments.
Skip it if you are:
- On Intel Mac, Windows, or Linux — it literally will not run.
- Looking for a team transcription platform with collaboration features.
- Needing broadcast-grade features like speaker labels, timestamps, or export formats for editing suites — grab Deepgram or a dedicated transcription tool instead.
- Wanting voice generation or cloning — that's Murf AI or ElevenLabs territory, not this.
- Uncomfortable buying from a very new indie tool with limited history.
Verdict
Hitoku Draft is a well-scoped niche tool that does one thing and mostly does it well. The $5 price makes it an easy yes for the target user — a privacy-conscious Apple Silicon Mac owner who wants local dictation without a subscription. The privacy story is real, the local inference is fast enough, and the one-time pricing is refreshing in a market that's gone subscription-crazy.
But the limitations are equally real. Apple Silicon exclusivity cuts out a huge chunk of potential users. The product is early-stage, which means fewer polish hours and more risk of the developer moving on. And the feature scope is deliberately narrow — this is not a Descript or Deepgram replacement.
My rating: 6.8/10. That number reflects a solid product held back by platform lock-in and early-stage maturity, not by execution.
Recommendation: If you're on an Apple Silicon Mac and you care about local, private voice-to-text, Hitoku Draft is worth the $5. Treat it like a well-priced utility rather than a critical piece of infrastructure, and you'll be happy with it. If you're on any other platform, or you need team features, look elsewhere — this isn't the tool for you and no amount of enthusiasm will change that.