voxmelt vs superwhisper

Superwhisper for Windows, or Voxmelt? Local and cloud, or local only.

Choosing between an app that can run local or cloud models and one that runs only on your own PC? Here is where the audio goes on Windows, whether each works offline, what each keeps, what each costs and how each rewrites text, with every Superwhisper fact linked to Superwhisper’s own site and the date we checked it.

the short answer

More devices and cloud models, or local only and one payment

If you dictate on a Mac or a phone as well as a PC, want cloud models, or need transcripts of audio and video files, Superwhisper covers more. If you want dictation and AI rewriting that only ever run on your Windows PC, with no cloud mode at all, and would rather pay once, $49 against $249.99 for Superwhisper’s one-time licence, that is what Voxmelt is for.

Sources: Superwhisper vs Wispr Flow, superwhisper.com; Choose a model and create your first mode, superwhisper.com; Home page, superwhisper.com. Checked on 26 October 2026.

Superwhisper

Local or cloud, on every device you own

Mac, Windows, iOS and Android. On Windows 10 and 11 each mode uses the local or cloud models you pick, and the defaults its setup guide recommends are cloud models. Free tier with unlimited use of its Whisper models; Pro is $8.49 a month, $84.99 a year or $249.99 once.

Sources: Superwhisper vs Wispr Flow, superwhisper.com; Superwhisper on Windows, superwhisper.com; Choose a model and create your first mode, superwhisper.com; Home page, superwhisper.com. Checked on 26 October 2026.

Voxmelt

Local only, on your Windows PC

Windows 10 and 11 only. Your speech is recognised on your own CPU or GPU and the AI rewrites on your own NVIDIA GPU, with no cloud mode, so your audio and transcripts never leave the PC. Free tier, then $49 or $99 once, with no subscription.

the free route first

What Windows already gives you

Both come with Windows, and Microsoft lists no separate price for either. If one of them does what you need, you need neither Superwhisper nor Voxmelt.

Voice typing (Win+H)

Start

Press Windows key + H on a hardware keyboard.[12]

Where your audio goes

It uses online speech recognition powered by Azure Speech services, so you need to be connected to the internet.[12]

Fluid dictation

On Copilot+ PCs, on-device small language models make voice typing smoother and smarter, in all English locales.[12]

Windows

Windows 11 and Windows 10.[12]

Voice access

What it does

Controls the PC and writes text with your voice.[13]

Where your audio goes

On-device speech recognition that works even without the internet; the first setup needs a connection to download the language files.[13][14]

Windows

Windows 11, version 22H2 and later, so not Windows 10.[13]

The language lists, and what Microsoft says about where your audio goes, are on Voxmelt vs Windows voice access and Win+H.

side by side

What each one does on Windows, checked against Superwhisper’s own pages

VoxmeltSuperwhisper
Windows and hardwareWindows 10 (64-bit) and Windows 11 on an x64 CPU. No graphics card to dictate; an NVIDIA GPU with 6 GB or more (Ampere or newer) for the AI text studioWindows 10 and 11 on x64 and ARM64; a GPU is not required. Its Cohere Transcribe and Whisper Large v3 Turbo models use the GPU through Vulkan when your GPU and drivers support it, and otherwise fall back to the CPU, which it calls much slower[2]
Other platformsNone: Windows onlymacOS, iOS and Android; one purchase covers Mac and Windows[4][2]
Where your audio goesStays on your PC: recognised on your own CPU or GPU, with no cloud modeEach mode uses the models you pick: local ones on your device, or cloud ones. The defaults its setup guide recommends, S1-Voice and S1-Language, are cloud models[5][7]
Works offlineYes, after the model download at setup; a paid licence keeps working offline for 30 days after its last successful checkYes on local models; cloud models require the internet. The local language model on its list, S1-mini, is English only and runs only the Message and Mail modes[5][7]
What the company keepsNo audio or text on a server. A hashed device fingerprint for the trial and licence check, your email only if you create the optional account, and payment records if you buy; the app has no telemetry, analytics or crash reportingYour email, licence key and purchase records; crash and error reports for up to 90 days and server logs for 30 days. Audio and transcripts from its cloud models are not stored on its servers after the request completes, except problem reports you send and meeting summaries you share; your history stays on your device[11][10]
Free plan15-day trial of everything, then 60 minutes of dictation a day; no cardA free tier that does not expire, with unlimited use of its Whisper models, and a 3,000-word trial of Pro[3][4]
Paid plansPro $49 or Studio $99, paid once; no subscriptionPro $8.49 a month or $84.99 a year, or a one-time licence at $249.99[4]
RefundWithin 7 days of purchase if the licence has not been activated; after that only for a verified defect we cannot fixWithin 30 days of purchase, no questions asked[3]
Computers per licenceFree tier and trial on 1 computer, Pro on up to 3, Studio with no device capAs many of your devices as you like[3]
AI rewritingOn your PC only, through Ollama on an NVIDIA GPU: 17 templates across 82 tones, plus your own templatesBuilt-in and custom modes with your own prompts, on cloud models from Superwhisper, OpenAI, Anthropic, Google or Groq, or on its local S1-mini[8][7]
Audio and video filesNo: Voxmelt dictates from your microphone and keeps no audio filePro transcribes audio and video files[3]
Published accuracy7.8% word error rate with Whisper large-v3 on an NVIDIA GeForce RTX 3080 Ti (12 GB), across 5 recorded clips, published 16 July 2026From its own benchmark: 6.6% word error rate for its cloud S1-Voice model and 8.4% for Cohere Transcribe, its most accurate local model; not the same test[5][9]
Security certificationsNone; Voxmelt makes no compliance claimSays it is SOC 2 Type II and HIPAA compliant[1]

Each bracketed number links to Superwhisper’s page for that fact; the full list is under Sources. Checked on 26 October 2026. Voxmelt’s accuracy comes from our published benchmark, which did not test Superwhisper, its prices from the pricing page, and what can leave your PC from our privacy policy.

the honest part

When Superwhisper is the better pick

If one of these is you, use Superwhisper. We would rather lose the sale than the argument.

You also dictate on a Mac, an iPhone or an Android phone

Superwhisper runs on macOS, Windows, iOS and Android, one purchase covers Mac and Windows, and you can activate your licence on as many of your devices as you like. Voxmelt is Windows only, with no port we are willing to put a date on.

Sources: Superwhisper vs Wispr Flow, superwhisper.com; Superwhisper on Windows, superwhisper.com; Home page, superwhisper.com. Checked on 26 October 2026.

Your Windows laptop has an ARM processor

Superwhisper runs natively on Windows on ARM64, where Parakeet Multilanguage is its local voice model. Voxmelt’s system requirements ask for an x64 CPU.

Sources: Superwhisper on Windows, superwhisper.com; Models: Voice, superwhisper.com. Checked on 26 October 2026.

You need transcripts of audio or video files, or of meetings

Superwhisper Pro transcribes audio and video files, and on Windows its guide points meetings to the Meeting mode with system audio. Voxmelt dictates only from your microphone and keeps no audio file.

Sources: Home page, superwhisper.com; Superwhisper on Windows, superwhisper.com. Checked on 26 October 2026.

You want AI formatting without an NVIDIA graphics card

Superwhisper’s cloud models run on its hosted servers, so they work on any machine, and your text goes to the cloud for that step. Voxmelt’s dictation runs on any modern CPU, but its AI rewriting needs an NVIDIA GPU with 6 GB or more.

Sources: Superwhisper on Windows, superwhisper.com; Models: Language, superwhisper.com. Checked on 26 October 2026.

Your company needs a certified vendor

Superwhisper’s Windows page says it is SOC 2 Type II and HIPAA compliant, and it offers an Enterprise plan. Voxmelt has no team plan, no admin console and no certification, and makes no compliance claim.

Sources: Superwhisper for Windows, superwhisper.com; Home page, superwhisper.com. Checked on 26 October 2026.

You want a long refund window

Superwhisper says its no questions asked refund within 30 days of purchase will always be honored. Voxmelt refunds within 7 days of purchase only if the licence has not been activated, which is why its trial is 15 days of everything with no card.

Source: Home page, superwhisper.com. Checked on 26 October 2026.

what voxmelt adds

What you get with Voxmelt

No cloud mode at all

Speech is recognised on your own CPU or GPU and the AI rewrites on your own NVIDIA GPU, so your audio and transcripts never leave the PC, whichever template you use. What does go online is the trial and licence check, the optional account if you make one, billing if you buy, update checks and model downloads, and none of it carries your audio or text.

Dictation with the Wi-Fi off

After the model download at setup, Voxmelt dictates with no internet connection, and its AI rewriting runs offline on your own GPU too. A paid licence keeps working offline for 30 days after its last successful check.

One payment, no subscription

Pro is $49 once and Studio $99 once, against $249.99 for Superwhisper’s one-time licence. Voxmelt Pro costs less than 6 months of Superwhisper Pro at $8.49 a month, and less than one year of it at $84.99. Before you pay, a 15-day trial of everything, then a free tier of 60 minutes of dictation a day, no card.

Source: Superwhisper vs Wispr Flow, superwhisper.com. Checked on 26 October 2026.

An AI text studio on your own GPU

With an NVIDIA GPU of 6 GB or more, a local language model served by Ollama rewrites raw dictation into an email, a summary, a commit message or a prompt: 17 templates across 82 tones, plus templates you write yourself.

Read-back and voiceover

On-device text-to-speech reads any text back in a natural voice and exports it as a WAV file (Pro), offline. Proofread by ear before you send.

An accuracy figure with its conditions

7.8% word error rate with Whisper large-v3 on an NVIDIA GeForce RTX 3080 Ti (12 GB), across 5 recorded clips, published on 16 July 2026 with the method and every clip, including the ones we lose on.

What it does not do: run on a Mac or a phone, transcribe audio or video files or meetings, or rewrite text with AI without an NVIDIA card. The full comparison lists where other tools beat it.

questions

Superwhisper and Voxmelt, answered

Does Superwhisper work on Windows?

Yes. Superwhisper says it runs natively on Windows 10 and 11, on x64 and ARM64, that a GPU is not required, and that it is the same app and the same licence as on macOS. Without a Vulkan-capable GPU, its Cohere Transcribe and Whisper Large v3 Turbo models fall back to the CPU, which it calls much slower; Parakeet Multilanguage does not need Vulkan. Voxmelt runs only on Windows 10 (64-bit) and Windows 11, and its default speech engine runs on the CPU.

Source: Superwhisper on Windows, superwhisper.com. Checked on 26 October 2026.

Does Superwhisper work offline?

With local models, yes. Its model guide says local models work offline and cloud models require the internet. Its Whisper models are free and run on your device; Cohere Transcribe and Parakeet are Pro. For AI formatting offline, its model list has S1-mini, a local language model that is English only and runs only the Message and Mail modes. Voxmelt needs the internet at setup, to download its speech model; after that it dictates with the Wi-Fi off, its AI rewriting runs on your own NVIDIA GPU, and a paid licence keeps working offline for 30 days after its last successful check.

Sources: Choose a model and create your first mode, superwhisper.com; Models: Voice, superwhisper.com; Models: Language, superwhisper.com. Checked on 26 October 2026.

Does Superwhisper send my voice to the cloud?

Only when a mode uses a cloud model, and the defaults its setup guide recommends, S1-Voice for transcription and S1-Language for formatting, are both cloud models. Its model guide says cloud audio is processed under a zero data retention policy. Its privacy policy says audio and transcripts from its cloud models are not stored on its servers after the request completes, except problem reports you choose to send and meeting summaries you choose to share, and that your data is not used for training AI models. With a local model, it says, your audio and transcripts never leave your device. In Super Mode the AI is also given text from the app you are in, text you have selected or copied within 3 seconds of dictating, and the date and time, your full name and your computer name; History shows everything that was sent to the AI. Voxmelt has no cloud mode: your audio and transcripts never leave the PC.

Sources: Choose a model and create your first mode, superwhisper.com; Models: Language, superwhisper.com; Privacy Policy, superwhisper.com; Built-in Modes, superwhisper.com. Checked on 26 October 2026.

How much does Superwhisper cost compared with Voxmelt?

Superwhisper has a free tier that does not expire, with unlimited use of its Whisper models, and you can try Pro for 3,000 words. Pro is $8.49 a month or $84.99 a year, and there is a one-time licence at $249.99; it says its no questions asked refund within 30 days of purchase will always be honored. Voxmelt starts with a 15-day trial of everything and then a free tier of 60 minutes of dictation a day, with no card; Pro is $49 once and Studio $99 once, refundable within 7 days of purchase if the licence has not been activated. Voxmelt Pro costs less than 6 months of Superwhisper Pro at $8.49 a month, and less than one year of it at $84.99.

Sources: Home page, superwhisper.com; Superwhisper vs Wispr Flow, superwhisper.com. Checked on 26 October 2026.

Which is more accurate, Superwhisper or Voxmelt?

There is no like-for-like answer. Superwhisper’s model guide gives a 6.6% word error rate for its cloud S1-Voice model and 8.4% for Cohere Transcribe, its most accurate local model, from its own benchmark of 16 speech models over 8 datasets, with local models run on Apple M4 and M4 Max hardware. Voxmelt publishes a 7.8% word error rate with Whisper large-v3 on an NVIDIA GeForce RTX 3080 Ti (12 GB), across 5 recorded clips, dated 16 July 2026; the clips cover clean dictation, technical / code speech, fast natural speech with filler, medical jargon and background noise, and the method is on our benchmark page. The tests are not the same, and our benchmark did not test Superwhisper. The fairest test is your own voice: dictate the same paragraph into each and count the fixes.

Sources: Choose a model and create your first mode, superwhisper.com; How We Benchmark, superwhisper.com. Checked on 26 October 2026.

Can I use one licence on more than one computer?

Superwhisper says you can activate your licence on as many of your devices as you like, and that one purchase covers Mac and Windows. Voxmelt’s free tier and trial run on 1 computer, Pro on up to 3 computers and Studio with no device cap; if you replace a computer and hit the cap, email help@voxmelt.com and we move the licence.

Sources: Home page, superwhisper.com; Superwhisper on Windows, superwhisper.com. Checked on 26 October 2026.

Does Voxmelt have SOC 2 or HIPAA, like Superwhisper?

No. Superwhisper’s Windows page says it is SOC 2 Type II and HIPAA compliant. Voxmelt holds no certification and makes no compliance claim. It processes your audio and text on your own PC and sends none of it to us, so there is no Voxmelt-side processor to sign an agreement with. That is a property of local processing, not a legal promise: your own obligations still apply to what you store, sync and share.

Source: Superwhisper for Windows, superwhisper.com. Checked on 26 October 2026.

Try it next to Superwhisper

Install both on the same PC, dictate the same paragraph into each, then turn the Wi-Fi off and try again. Free tier with no card, a 15-day trial of everything, and $49 once after that.

Direct installer, for PCs without the Store; Windows will warn because it is not code-signed.

Read how offline dictation works on Windows and the free routes to private dictation on Windows 11, or see Voxmelt against Wispr Flow, Superwhisper, Otter AI and Dragon.

sources

Where every Superwhisper and Windows fact comes from

Superwhisper’s own pages on superwhisper.com and Microsoft’s own support pages, read on the date given. If one of them has changed since, tell us at help@voxmelt.com and we will fix this page.

  1. Superwhisper for Windows, superwhisper.com. Checked on 26 October 2026.
  2. Superwhisper on Windows, superwhisper.com. Checked on 26 October 2026.
  3. Home page, superwhisper.com. Checked on 26 October 2026.
  4. Superwhisper vs Wispr Flow, superwhisper.com. Checked on 26 October 2026.
  5. Choose a model and create your first mode, superwhisper.com. Checked on 26 October 2026.
  6. Models: Voice, superwhisper.com. Checked on 26 October 2026.
  7. Models: Language, superwhisper.com. Checked on 26 October 2026.
  8. Built-in Modes, superwhisper.com. Checked on 26 October 2026.
  9. How We Benchmark, superwhisper.com. Checked on 26 October 2026.
  10. Sensitive Data Best Practices, superwhisper.com. Checked on 26 October 2026.
  11. Privacy Policy, superwhisper.com. Checked on 26 October 2026.
  12. Use voice typing to talk instead of type on your PC, Microsoft Support. Checked on 26 October 2026.
  13. Use voice access to control your PC & author text with your voice, Microsoft Support. Checked on 26 October 2026.
  14. Set up voice access, Microsoft Support. Checked on 26 October 2026.

Voxmelt’s figures come from our benchmark, published 16 July 2026, and our pricing page, and what can leave your PC is listed in our privacy policy.