Direct answer — Is VoiceInk worth it in 2026? Yes, if you use an Apple Silicon Mac and want open-source dictation. VoiceInk’s code is free under GPLv3, so you can build it yourself, and the ready-made app costs $25 to $49 once, depending on how many Macs you cover. The trade-offs: no Intel or Windows support, self-built copies lose automatic updates, and its accuracy figure is the developer’s own claim.
This VoiceInk review looks at one of the best-known open-source dictation apps for the Mac. VoiceInk uses local AI models to turn what you say into text in any app, and its source code sits on GitHub for anyone to read, audit or build.
We checked every claim below against VoiceInk’s website, pricing page, FAQ, model documentation and GitHub repository on 7 November 2026. We did not run our own speed or accuracy tests. Where VoiceInk does not publish something, we say so. At the end, we compare it with Weesper Neon Flow.
What is VoiceInk?
VoiceInk is a native macOS dictation app that transcribes your speech and inserts the text where you are typing. Its GitHub README describes it as an app that turns speech into text “almost instantly”, using local AI models.
The project is maintained by an independent developer known as Pax, who handles support and the roadmap through GitHub issues and email, according to the FAQ. On 7 November 2026, the repository showed about 7,000 stars and more than 1,900 commits.
What VoiceInk lists among its features, across its homepage and README:
- Custom words: names and product terms recognised as you spell them
- Smart Replace: short phrases expanded into full text, or one word swapped for another
- Modes: presets that switch settings per app or per website
- Context awareness: the app can use what is on screen to improve results
- Text cleanup and AI writing modes: filler words removed, emails formatted, selected text rewritten on a spoken instruction
- Shortcuts: keyboard or mouse triggers, including push-to-talk
Is VoiceInk really free and open source?
Yes for the code, no for the ready-made app. The README states that the project is licensed under the GNU General Public License v3.0, and the LICENSE file is the standard GPL version 3 text dated 29 June 2007.
GPLv3 lets you run, study, modify and share the software, provided that what you distribute stays under the same licence. The project does not accept pull requests, but you can fork it.
What does it take to build VoiceInk yourself?
VoiceInk’s build guide (BUILDING.md) lists three requirements:
- macOS 15.0 or later on the build machine
- Xcode with Swift 6.2 or later and the Command Line Tools
- Git to clone the repository
You then run make local, which prepares the Whisper framework, builds a release version and copies VoiceInk.app to your Downloads folder. Ad-hoc signed builds may need macOS permissions granted again after each rebuild.
The guide is clear about the cost of doing it yourself: “Local builds do not include iCloud dictionary sync or automatic updates.” The README adds that a paid licence gives you automatic updates, priority support and upcoming features, and helps fund development.
For more free and open-source options, our guide to the best free offline dictation apps compares them side by side.
How much does VoiceInk cost?
VoiceInk sells three one-time licences, shown in dollars on its pricing page on 7 November 2026. Each includes lifetime updates and a 14-day money-back guarantee.
| Plan | Price | Macs covered |
|---|---|---|
| Solo | $25 once | 1 |
| Personal | $39 once | Up to 2 |
| Extended | $49 once | Up to 3 |
A few points worth knowing before you buy:
- No subscription: the page states “No subscriptions, no recurring fees.”
- Licence key: you receive a key after purchase, and you manage devices through the Polar customer portal.
- Trial: the site offers a “Download Free Trial” button, but neither the pricing page nor the FAQ states how long the trial lasts.
Prices change, so check the vendor’s page on the day you buy. For context on one-time versus monthly plans across the market, see our voice dictation pricing comparison.
Does VoiceInk work offline, and how private is it?
Mostly, if you stay on local models. VoiceInk’s FAQ says it “processes all voice transcription locally on your device” and that your voice data never leaves your Mac. Local models run on the Mac’s Neural Engine.
Two optional features change that picture:
- Cloud transcription providers: the recommended models page lists ElevenLabs Scribe V2, Deepgram Nova 3, Soniox, Groq Whisper Large v3 Turbo and AssemblyAI. A cloud provider transcribes audio on its own servers, so its own privacy terms then apply.
- Cloud Enhancement: the FAQ says this optional feature processes the transcribed text, not your voice.
The open-source code adds something closed apps cannot offer: you can read it and check what the app sends.
Which models and languages does VoiceInk support?
VoiceInk lets you pick between several local models rather than shipping one. Its recommended models page, checked on 7 November 2026, lists:
| Local model | What VoiceInk says it covers |
|---|---|
| Parakeet Ultra | First pick on Apple Silicon; English and supported European languages |
| SenseVoice Small | Mandarin, Cantonese, Japanese, Korean and English |
| Cohere Transcribe | Multilingual, including Arabic and Vietnamese |
| Whisper Large v3 Turbo | Multilingual alternative; processes audio after you stop speaking |
VoiceInk does not publish a single list of supported languages. The documentation tells you to check each Mode’s language picker before downloading a model. If you want background on the two main model families, read our comparison of Parakeet and Whisper.
On accuracy, the README claims that local models reach “99% accuracy”, without naming a test set, a language or audio conditions. Treat it as a vendor claim, not a benchmark.
Which Macs does VoiceInk support?
VoiceInk requires an Apple Silicon Mac and macOS 14.4 or later, according to its FAQ and pricing page, with 8 GB of RAM recommended. Intel Macs are not supported, and the model documentation warns that local models do not work reliably on them.
One inconsistency is worth flagging: the GitHub README and the build guide state macOS 15.0 or later. If you run macOS 14, check the requirements of the release you download.
VoiceInk’s homepage announces an iOS version; there is no Windows or Linux version.
VoiceInk vs Weesper Neon Flow: how do they compare?
Both apps transcribe on your device and insert text at the cursor. They differ mainly in platform coverage, licensing model and source code.
| Weesper Neon Flow | VoiceInk | |
|---|---|---|
| Platforms | macOS 13+, Windows 10 (19041+) / 11 | macOS 14.4+ (README: 15.0+); iOS version announced |
| Mac processors | Apple Silicon, or Intel with AVX2 | Apple Silicon only |
| Source code | Closed | Open source, GPLv3 |
| Dictation languages (vendor claims) | 55+ | No single list published; depends on the model |
| Vocabulary and snippets | Custom dictionary and snippets | Custom words and Smart Replace |
| Cloud options | No cloud transcription | Optional cloud providers and Cloud Enhancement |
| Trial | 15 days, no credit card | Free trial offered; length not stated |
| After the trial | Free Lite: 60 seconds per dictation, 1,500 words per week | Build it yourself from source |
| Paid plans | €5/mo, €45/yr, €99 lifetime | See the pricing table above |
| Devices per licence | Up to three computers | 1, 2 or 3 Macs, depending on the plan |
Prices stay in each vendor’s own currency: euros for Weesper, dollars for VoiceInk. We do not convert them, because a converted price is one neither company charges.
Where VoiceInk is stronger: its code is open and auditable, you can build it for free, the paid licences are cheap one-time purchases, and it offers a choice of local models plus AI writing modes.
Where Weesper Neon Flow is stronger: it runs on Windows, on Intel Macs with AVX2 and on macOS 13 Ventura, so you keep one consistent dictation workflow across Mac and Windows. It has no cloud transcription option to switch off, and its trial lasts a stated 15 days. One honest caveat on our side: transcription never leaves your device, but the Weesper licence is verified online periodically and the app works offline for up to 72 hours between checks.
Want to compare them on your own writing? Start the free 15-day Weesper trial, no credit card required, and dictate the same email in both apps. For other head-to-heads, browse our dictation app comparisons.
How does VoiceInk compare with other open-source dictation apps?
VoiceInk is the most Mac-focused of the open-source dictation apps we have reviewed. Our Handy review covers a free cross-platform alternative, and our Amical review looks at an open-source app for Mac and Windows.
If you mainly transcribe files rather than dictate, our list of MacWhisper alternatives is a better starting point.
Who should choose VoiceInk?
Choose VoiceInk if you have an Apple Silicon Mac on a recent macOS, want software whose code you can audit, and are happy to pay a small one-time fee for updates or to build it yourself. The 14-day money-back guarantee keeps the risk of the paid licence low.
Look elsewhere if you use Windows or an Intel Mac, need the same tool on several operating systems, or prefer an app with no cloud options at all. In that case, download Weesper Neon Flow and try it for 15 days. Setup questions are answered in the Weesper Help Center.
Verdict
In our assessment, VoiceInk is one of the strongest open-source dictation apps on the Mac: actively developed, with a fresh release (v2.26) on GitHub on 7 November 2026, a choice of local models, and fair one-time pricing for those who do not want to build it.
Its limits are clear too: Apple Silicon only, slightly inconsistent macOS requirements between its website and its repository, an unstated trial length, and an accuracy figure measured by nobody but the developer. For a Mac-only user who values open code, it is well worth trying. For dictation across Mac and Windows, a cross-platform tool fits better.