Superwhisper vs Wispr Flow
Superwhisper and Wispr Flow both turn speech into text across apps, but the surrounding systems differ. Superwhisper offers on-device and cloud models, custom modes, file and meeting workflows, and a lifetime plan. Wispr Flow always transcribes in the cloud, reaches Android as well as desktop and iPhone, and combines 100+ languages with snippets, dictionaries, Command Mode, and team controls. IraVoice is the much smaller third option: on-device dictation with one held key on Mac, Windows, and Linux, code names spelled in terminals and coding agents, and a one-time license, with an optional Cloud plan coming soon.
Side by side
A factual comparison of the source-reviewed product boundaries. Follow the source links below for the latest competitor details.
| Factor | Superwhisper | Wispr Flow | IraVoice |
|---|---|---|---|
| Speech path | Offers on-device models that keep audio local and optional cloud models that send audio or text through Superwhisper’s proxy | Transcription always runs in Wispr’s cloud; Privacy Mode and Private Cloud Sync govern training and server storage separately | On-device NVIDIA Parakeet recognition after a one-time model download; Cloud recognition planned for IraVoice Cloud, coming soon |
| Platforms | Mac, Windows, iPhone, and iPad; one Pro license covers the supported personal devices | Mac, Windows, iPhone, and Android; iPad and Linux are not currently supported | macOS 13+ on Apple silicon, Windows 10 or 11 (64-bit), and Linux (beta); no mobile app |
| Price and trial | Free tier; Pro is $8.49 monthly, $84.99 yearly, or $249.99 lifetime according to current documentation | Basic is free with 2,000 words per week on desktop; Pro is $15 monthly or $12 per month billed annually after a 14-day trial | $79 once for IraVoice Local, after the 3-day free trial; IraVoice Cloud is coming soon |
| Languages | Advertises 100+ languages; exact support varies with the selected voice model | Advertises 100+ languages with session-level detection | English by default; other offline models cover 25+ languages; IraVoice Cloud, coming soon, is planned to add Cloud recognition in any language |
| Customization | Local and cloud voice models, built-in modes, custom AI instructions, app or website auto-activation, context, and optional provider keys | Dictionary and snippets on Basic; paid or trial access adds Command Mode, bulk import, and wider team controls | One dictation shortcut (hold or tap), custom words, replacements and snippets, and code names from your project; no custom-mode builder |
| Files and meetings | Documents audio and video file transcription, system-audio recording, meeting modes, and speaker separation | Documents a meeting notetaker; other-participant system audio requires macOS 14.4+, and Notetaker requires Private Cloud Sync | Does not transcribe imported media, record meetings, or identify speakers |
| Privacy and history | On-device models keep microphone audio local; cloud models use API paths with published zero-retention terms, while transcript history stays on the device | Cloud transcription; zero data retention requires Privacy Mode on and Private Cloud Sync off, while snippets and dictionary entries still sync | IraVoice Local sends no audio or text; model download and license check-ins are separate boundaries; IraVoice Cloud, coming soon, will process audio and text in its cloud, where IraVoice’s servers never store them, Groq does not retain them, Anthropic deletes API inputs and outputs within 30 days, and neither trains on them |
| Account and workflow | Offline mode can run without an account; Pro activation and cross-device licensing use the purchased license | Requires an account and is designed as a connected service across supported devices | No account; a license key is activated with a code emailed to the purchase address and works offline between weekly check-ins |
Decision guide
Primary sources
Competitor details were reviewed July 26, 2026. Pricing, platform support, and policies can change.
Current platforms, free and Pro positioning, pricing, languages, meetings, files, custom prompts, and model options.
Current monthly, annual, and lifetime prices, Free-versus-Pro boundaries, and cross-platform license coverage.
Current on-device and cloud voice models, tier boundaries, language coverage, proxy path, and processing disclosures.
Current built-in and custom modes, local and cloud model selection, auto-activation, system audio, meetings, and speaker separation.
Cloud-provider zero-retention terms, local history, FileSync boundary, dictionary storage, and compliance caveats.
Current Basic and Pro limits, prices, 14-day trial, platforms, languages, Command Mode, dictionary, snippets, and teams.
Current platform requirements, Mac permissions, meeting-notetaker system-audio boundary, and unsupported devices.
Current cloud-transcription path, training and storage controls, zero-retention configuration, sync exceptions, and Notetaker dependency.
Questions
Superwhisper documents on-device models that work without an internet connection. Wispr says transcription always runs in its cloud. IraVoice Local also runs speech recognition on the computer after a one-time model download, but its license check-ins and the optional IraVoice Cloud plan (coming soon) are separate connected boundaries.
The answer depends on usage and time. Superwhisper currently documents a free tier plus Pro at $8.49 monthly, $84.99 annually, or $249.99 lifetime. Wispr Flow Basic is free but capped at 2,000 words per week on desktop; Pro is $15 monthly or $12 per month billed annually. IraVoice Local is $79 once and IraVoice Cloud is coming soon, but IraVoice is substantially narrower than either product.
Both cover Mac, Windows, iPhone, and Android, and Superwhisper also documents iPad support. IraVoice runs on Apple-silicon Macs with macOS 13 or later, Windows 10 and 11, and Linux (beta), with no mobile app.
Superwhisper is the broader fit for model choice, custom modes, AI instructions, auto-activation, context, and provider keys. Wispr Flow offers dictionary entries and snippets, with paid Command Mode and team controls. IraVoice intentionally avoids a general custom-mode system; it offers custom words, replacements and snippets, and a recordable shortcut.
Superwhisper is the clearer documented fit for files, system audio, meeting modes, and speaker separation. Wispr Flow now documents a meeting notetaker whose system-audio capture requires macOS 14.4 or later and whose storage path requires Private Cloud Sync. IraVoice does neither job.
There is no honest universal winner without naming the active configuration. Superwhisper’s on-device models keep audio local, while its cloud models use a separate proxy and provider path. Wispr always transcribes in the cloud, with zero data retention requiring Privacy Mode on and Private Cloud Sync off. IraVoice Local keeps recognition on the computer but still has named model-download and license check-in connections, while IraVoice Cloud, coming soon, will send audio and text to its cloud, where IraVoice’s servers never store them, Groq does not retain them, Anthropic deletes API inputs and outputs within 30 days, and neither trains on them.
No. Superwhisper is broader for models, modes, files, meetings, and offline choices. Wispr Flow is broader for devices, languages, personalization, and teams. IraVoice is only the stronger fit when one held key on Mac, Windows, and Linux, a lower one-time price, and code names spelled in developer tools match the job.
macOS 13+ on Apple silicon, Windows 10 or 11 (64-bit), and Linux (beta)
Try IraVoice for 3 days with no email, card, or account, then choose it only if that narrow workflow beats Superwhisper’s configurability or Wispr Flow’s cloud reach.