Find the Dragon alternative that fits your work.

FluidVoice wins on ongoing app cost: local dictation and cleanup are free without a word limit or paid tier. VoiceInk offers a $25 one-time Mac license when you want a purchased app rather than building from source. Ottex offers both free local-model access and free BYOK app access without requiring a one-time license; your API provider charges separately, while Pro costs $14/month. Separate app purchases and subscriptions from model usage charges before comparing totals.

See the evidence — 25 claims across 6 toolsverified Oct 2026Show tableHide table

	Ottex	Dragon NaturallySpeaking	Spokenly	FluidVoice	Wispr Flow	Krisp
Free tier	YesSource: ottex.ai — checked Oct 2026Forever free with local models and BYOK providers.	NoNo free plan	PartialSource: spokenly.app/pricing — checked Oct 2026No usage limits	PartialSource: fluidvoice.org/download — checked Oct 2026no trial, word limit or paid tier	PartialSource: wisprflow.ai/pricing — checked Oct 20262,000/week on desktop; 1,000/week on mobile; Weekly limit	PartialSource: krisp.ai/blog — checked Oct 2026Unlimited transcription; AI noise cancellation (60 minutes daily); AI summaries & action items (2 daily); Meeting history (limited)
Subscription	$14/moSource: ottex.ai — checked Oct 2026$14/mo — unlimited dictation and meeting intelligence, all hosted models included.	NoNo subscription plan	$6.99/moSource: spokenly.app/team — checked Oct 2026Each seat includes full Pro with a permanent 30% discount and centralized billing.	NoNo subscription plan	$15/moSource: docs.wisprflow.ai/articles — checked Oct 2026Pro also serves teams at the same per-user prices without a minimum seat requirement; Notetaker access is rolling out gradually.	$16/moSource: krisp.ai/pricing — checked Oct 2026Monthly billing is $16/mo/user and annual billing is advertised as $8/mo/user with Save 50%, without a published annual total.
One-time payment	Not applicableSource: ottex.ai — checked Oct 2026Application is free	NoNo one-time payment plan	Not applicableSource: spokenly.app/pricing — checked Oct 2026Application is free	Not applicableSource: fluidvoice.org/download — checked Oct 2026Application is free	NoNo one-time payment plan	NoNo one-time payment plan
Bring your own keyPay the model provider, not the middleman	YesSource: ottex.ai — checked Oct 2026OpenRouter, Google AI, OpenAI, Anthropic, Mistral, and compatible providers.	No public claim	YesSource: spokenly.app/privacy — checked Oct 2026BYOK audio and text travel directly from the device to the selected provider.	YesSource: fluidvoice.org/docs — checked Oct 2026Optional cloud cleanup uses the user's API key and separately billed provider account.	No public claim	No public claim
Education discount	NoSource: ottex.ai — checked Oct 2026Same pricing for everyone.	YesSource: dragon.nuance.com/en-us — checked Oct 2026Qualified K–12 and higher-education organizations receive OLP discounts with unpublished prices, billing periods and percentages; the separate K–12 site program covers up to 250 computers.	YesSource: spokenly.app/students — checked Oct 2026Verified students receive 50% off Pro at $4.99/month or $49.99/year after signing in with an academic email and emailing a student ID to support.	No public claim	YesSource: docs.wisprflow.ai/articles — checked Oct 2026Eligible students can claim $0/year on desktop or purchase $7.50/month or $72/year, representing 50% off standard Pro; the separate student marketing page advertises three months free followed by $6/month billed annually, which conflicts with the plans documentation.	No public claim

Dragon supports Windows 10, Windows 11, and several Windows Server versions, but its desktop product has no Mac or Android app. Wispr Flow covers Mac, Windows, iOS, and Android through installed apps; its browser demo does not replace them. Ottex covers Mac and iOS only. For this decision, shipped operating-system coverage is clearer than a native-versus-Electron comparison, since the application frameworks are not specified.

See the evidence — 9 claims across 6 toolsverified Oct 2026Show tableHide table

	Ottex	Dragon NaturallySpeaking	Spokenly	FluidVoice	Wispr Flow	Krisp
Platforms	Mac, iOSSource: ottex.ai — checked Oct 2026	WindowsSource: dragon.nuance.com/shared — checked Oct 2026	Mac, Windows, iOS, LinuxSource: spokenly.app/download — checked Oct 2026Some platforms in beta/waitlist	MacSource: fluidvoice.org/download — checked Oct 2026Some platforms in beta/waitlist	Mac, Windows, iOS, AndroidSource: docs.wisprflow.ai/articles — checked Oct 2026	Mac, Windows, iOS, Android, WebSource: help.krisp.ai/hc — checked Oct 2026
Native app footprintLight on memory, instant to invoke	YesSource: ottex.ai — checked Oct 2026Native Swift app, ~50MB RAM.	No public claim	YesSource: spokenly.app/android — checked Oct 2026The macOS and iOS applications are built with Swift.	PartialSource: raw.githubusercontent.com/altic-dev — checked Oct 2026The macOS application uses SwiftUI and AppKit; the Windows implementation was not verified.	No public claim	No public claim

Dragon's single-speaker file transcription cannot replace live meeting capture or speaker separation. Ottex captures system audio without a bot, separates speakers, and searches across meetings, but its public share links are not a full team workspace. Krisp Core costs $16/seat/month and adds audio and screen recording, automatic speaker names, and shared meeting contexts. Ottex's camera recording watermark is experimental and requires setup; its participant chat/calendar recording notices remain unconfirmed.

See the evidence — 38 claims across 6 toolsverified Oct 2026Show tableHide table

	Ottex	Dragon NaturallySpeaking	Spokenly	FluidVoice	Wispr Flow	Krisp
Meeting recordingThe call is captured at all	YesSource: ottex.ai — checked Oct 2026With speaker separation.	NoSource: dragon.nuance.com/en-us — checked Oct 2026Not supported.	PartialSource: spokenly.app/dictation-for-mac — checked Oct 2026Meeting recording ships on Mac and requires macOS 14.4 or later.	PartialSource: github.com/altic-dev — checked Oct 2026FluidMeet records meetings locally in macOS v1.6.10-beta.7.	PartialSource: docs.wisprflow.ai/articles — checked Oct 2026Notetaker records on Mac and Windows, with account rollout, organization, and HIPAA restrictions.	YesSource: help.krisp.ai/hc — checked Oct 2026Core and Advanced support transcription-only, audio recording and screen recording modes.
Bot-free captureNo "recording bot" joins your call	YesSource: ottex.ai — checked Oct 2026System-audio capture, no bot joins the call.	NoSource: dragon.nuance.com/en-us — checked Oct 2026Not supported.	YesSource: spokenly.app/blog — checked Oct 2026Mac meeting recording captures audio without a bot joining the call.	PartialSource: raw.githubusercontent.com/altic-dev — checked Oct 2026The macOS beta captures meeting-application audio and microphone audio directly on the computer.	YesSource: docs.wisprflow.ai/articles — checked Oct 2026Notetaker captures audio from the user's device without joining the call as a bot.	YesSource: krisp.ai/ai-meeting-assistant — checked Oct 2026Desktop recording can capture meetings without a bot joining the call.
Camera recording watermarkShow recording in your camera video	PartialExperimental macOS virtual camera adds an Ottex icon to video; install/enable the extension, select it and turn video on. Guide covers Meet in Chrome and Zoom desktop; effects may obscure it.	NoSource: dragon.nuance.com/en-us — checked Oct 2026Not supported.	No public claim	No public claim	No public claim	No public claim
Participant noticesRecording notices in chat or calendar	No public claim	NoSource: dragon.nuance.com/en-us — checked Oct 2026Not supported.	No public claim	No public claim	PartialSource: docs.wisprflow.ai/articles — checked Oct 2026Mac and Windows can post chat notices in Zoom, Teams, Slack huddles, and foreground Chrome Google Meet; Google Calendar notices require organizer write access, while organization policy can lock chat notices on.	PartialSource: help.krisp.ai/hc — checked Oct 2026Notify others you're using Krisp enables chat or audio notices, with possible platform blocking and no verified calendar-invite notice or affirmative consent collection.
Speaker separation	YesSource: ottex.ai — checked Oct 2026	NoSource: dragon.nuance.com/en-us — checked Oct 2026Not supported.	PartialSource: spokenly.app/blog — checked Oct 2026Saved meeting recordings can be transcribed with voice-based speaker labels that may require manual correction.	PartialSource: github.com/altic-dev — checked Oct 2026FluidMeet includes speaker labels in the macOS beta; automatic identification of participants by real name is unverified.	PartialSource: docs.wisprflow.ai/articles — checked Oct 2026Names use calendar details, visible participant tiles, and spoken introductions; unsupported apps, hidden tiles, or shared microphones can retain numbered labels.	YesSource: linkedin.com/pulse — checked Oct 2026Krisp automatically applies speaker names in transcripts and notes while allowing manual edits.
Calendar detection	YesSource: ottex.ai — checked Oct 2026Calendar-based meeting detection.	NoSource: dragon.nuance.com/en-us — checked Oct 2026Not supported.	PartialSource: spokenly.app/releases — checked Oct 2026Mac meeting detection is documented in release notes, including Safari support.	PartialSource: raw.githubusercontent.com/altic-dev — checked Oct 2026The macOS beta optionally detects Zoom, Teams, Webex, and supported browser meetings, then offers user-started recording.	YesSource: docs.wisprflow.ai/articles — checked Oct 2026Notetaker detects calls and presents a recording card, with configurable automatic starting.	YesSource: help.krisp.ai/hc — checked Oct 2026With Note Taker enabled and audio devices configured, Krisp captures the next call automatically.
Cross-meeting search	YesSource: ottex.ai — checked Oct 2026	NoSource: dragon.nuance.com/en-us — checked Oct 2026Not supported.	No public claim	PartialSource: github.com/altic-dev — checked Oct 2026The macOS beta's Global Search includes transcripts across the application.	YesSource: docs.wisprflow.ai/articles — checked Oct 2026Ask Wispr in the Notetaker hub searches across meeting notes and cites source meetings.	YesSource: linkedin.com/pulse — checked Oct 2026Global AI Chat answers questions across meetings and identifies which conversations informed the answer.
Shared team memory	PartialSource: ottex.ai — checked Oct 2026Public share links, not a full team workspace.	NoSource: dragon.nuance.com/en-us — checked Oct 2026Not supported.	No public claim	No public claim	PartialSource: docs.wisprflow.ai/articles — checked Oct 2026Shared notes appear in search and Ask Wispr context, subject to sharing permissions and separate transcript-access restrictions.	YesSource: help.krisp.ai/hc — checked Oct 2026AI Chat supports All Meetings, folders and Shared with Me contexts containing accessible meeting content.

Ottex drafts follow-up emails, CRM updates, and Linear tickets, with rough notes guiding the result. Its meeting template library structures transcript information into documents such as product reviews, while its separate transcript chat prompt library supplies reusable questions and analysis prompts. Krisp Core at $16/seat/month also offers both libraries and incorporates manual notes, but its HubSpot note delivery and internal action items are distinct from structured CRM-update drafts and Linear or Jira ticket drafts. Neither type of meeting library is an automatic dictation transformation.

See the evidence — 23 claims across 6 toolsverified Oct 2026Show tableHide table

	Ottex	Dragon NaturallySpeaking	Spokenly	FluidVoice	Wispr Flow	Krisp
Meeting template libraryTurn a transcript into a structured document	YesSource: ottex.ai — checked Oct 2026Choose a ready-made document template to extract relevant information from a meeting transcript into a structured document, such as a product review or client-call report.	NoSource: dragon.nuance.com/en-us — checked Oct 2026Document templates that extract and structure information from meeting transcripts are not supported.	No public claim	No public claim	No public claim	YesSource: help.krisp.ai/hc — checked Oct 2026Ready-made templates such as Project Sync, Client Status Update and Hiring generate structured documents from transcripts and can be selected per meeting or set as the default.
Transcript chat prompt libraryReady-made prompts for AI chat about meetings	YesSource: ottex.ai — checked Oct 2026Use ready-made reusable prompts in AI chat to ask questions, analyze discussions and extract information from meeting transcripts.	NoSource: dragon.nuance.com/en-us — checked Oct 2026Reusable AI chat prompts for analyzing meeting transcripts are not supported.	No public claim	No public claim	No public claim	YesSource: help.krisp.ai/hc — checked Oct 2026Core and Advanced users can select reusable suggested prompts such as Summarize call by speakers and List Q&A in a meeting's Ask Anything tab.
Decisions, owners, action itemsLeave the meeting with next steps clear	YesSource: ottex.ai — checked Oct 2026Decisions with owners extracted.	NoSource: dragon.nuance.com/en-us — checked Oct 2026Not supported.	No public claim	PartialSource: raw.githubusercontent.com/altic-dev — checked Oct 2026The macOS beta offers a Key decisions output, while assignment of decision owners is unverified.	YesSource: docs.wisprflow.ai/articles — checked Oct 2026Generated summaries include a Decisions section with speaker labels.	YesSource: krisp.ai/ai-meeting-assistant — checked Oct 2026Meeting notes identify key decisions and next steps, and action items can be assigned to participants.
Follow-up email draftedThe outcome is sent, not summarized	YesSource: ottex.ai — checked Oct 2026Drafted from templates or your own.	NoSource: dragon.nuance.com/en-us — checked Oct 2026Not supported.	No public claim	No public claim	PartialSource: wisprflow.ai/notetaker — checked Oct 2026Connected AI tools can draft follow-ups from meeting history through MCP; a dedicated built-in follow-up-email action is unverified.	PartialSource: help.krisp.ai/hc — checked Oct 2026Krisp can email meeting recaps to invitees, but a dedicated editable follow-up email drafting workflow is unverified.
CRM update / ticket draft	YesSource: ottex.ai — checked Oct 2026Structured for the target app.	NoSource: dragon.nuance.com/en-us — checked Oct 2026Not supported.	No public claim	No public claim	No public claim	PartialSource: help.krisp.ai/hc — checked Oct 2026Meeting notes can be sent automatically or manually to a selected HubSpot contact, company or deal, without a verified CRM-field draft review workflow.
Your rough notes steer the result	YesSource: ottex.ai — checked Oct 2026Rough notes and instructions steer the result.	NoSource: dragon.nuance.com/en-us — checked Oct 2026Not supported.	NoSource: spokenly.app/comparison — checked Oct 2026Not supported.	No public claim	PartialSource: docs.wisprflow.ai/articles — checked Oct 2026Ask Wispr reads typed meeting notes as chat context; automatic summaries are stored separately, and note-guided automatic summary generation is unverified.	YesSource: linkedin.com/pulse — checked Oct 2026Users' manual notes are incorporated into generated summaries, with visual cues distinguishing their contributions from AI content.

Dragon already offers local recognition, trained vocabulary, and voice commands; replacing it merely to get offline speech recognition changes little. Spokenly adds selectable models and a Local Only Mode that blocks outbound network access, with unlimited free local dictation and saved writing instructions. Ottex offers on-device Whisper and Parakeet v3, model selection by task and provider, 100 languages, mid-sentence language switching, and per-app rules that can preserve exact wording in AI chats while polishing emails.

See the evidence — 54 claims across 6 toolsverified Oct 2026Show tableHide table

	Ottex	Dragon NaturallySpeaking	Spokenly	FluidVoice	Wispr Flow	Krisp
Engines	Cloud: Whisper, Gemini 3 Flash, Voxtral. Local: Whisper, Parakeet v3.Source: ottex.ai — checked Oct 2026Hosted routing, BYOK, or fully local.	Nuance Dragon speech engineSource: dragon.nuance.com/shared — checked Oct 2026The desktop recognition engine uses Nuance Deep Learning technology.	Whisper, NVIDIA Parakeet, Apple Speech Analyzer, OpenAI GPT-4o Transcribe, Deepgram Nova, SonioxSource: spokenly.app/docs — checked Oct 2026Local and cloud transcription engines are selectable per mode.	Nemotron Speech 3.5, Nemotron 3.5 Multilingual, Parakeet Flash Beta, Parakeet TDT v3, Parakeet TDT v2, Cohere Transcribe, Apple Speech, WhisperSource: fluidvoice.org/models — checked Oct 2026The macOS model lineup varies by hardware and operating-system compatibility.	Cloud speech recognition with proprietary post-processingSource: wisprflow.ai/why-flow — checked Oct 2026Wispr describes proprietary models layered over speech recognition and introduced its Canto speech model in September 2026.	No public claim
Local / on-device optionAudio never leaves the machine	YesSource: ottex.ai — checked Oct 2026Whisper and Parakeet v3 run fully on-device.	YesSource: nuance.com/asset — checked Oct 2026Local recognition supports disconnected operation, subject to administrative configuration in managed deployments.	YesSource: spokenly.app/docs — checked Oct 2026Downloaded local models transcribe offline, and Local Only Mode blocks outbound network access.	YesSource: fluidvoice.org — checked Oct 2026Local recognition and local cleanup work offline after their required assets are installed.	NoSource: docs.wisprflow.ai/articles — checked Oct 2026Dictation transcription requires an internet connection and runs in the cloud.	NoSource: help.krisp.ai/hc — checked Oct 2026Not supported for system-wide dictation.
Model selection per task	YesSource: ottex.ai — checked Oct 2026Choose the model per task and per provider.	No public claim	YesSource: spokenly.app/docs — checked Oct 2026Users choose transcription models and AI providers per mode.	YesSource: fluidvoice.org/models — checked Oct 2026Users can download and switch speech models in the application.	No public claim	NoSource: help.krisp.ai/hc — checked Oct 2026Not supported for system-wide dictation.
Languages & mixed speechWork that crosses languages mid-sentence	YesSource: ottex.ai — checked Oct 2026100+ languages · Mid-sentence language switching.	No public claim	PartialSource: spokenly.app/blog — checked Oct 2026100+ languages · Gemini 3.5 Transcribe supports switching languages mid-sentence.	PartialSource: fluidvoice.org/docs — checked Oct 202699+ languages · Mid-sentence language switching can produce errors and requires testing with the selected model.	PartialSource: docs.wisprflow.ai/articles — checked Oct 2026100+ languages · Mixed Hindi-English speech is supported through explicitly selected Hinglish, and Polish preserves language blends while detection chooses one language per dictation.	NoSource: help.krisp.ai/hc — checked Oct 2026Not supported for system-wide dictation.
Automatic formatting	YesSource: ottex.ai/docs — checked Oct 2026Dictation automatically fixes punctuation, groups ideas into paragraphs and formats enumerations as lists. Formatting toggles and profile rules customize output.	PartialSource: nuance.com/products — checked Oct 2026Configurable rules format numbers, dates and addresses and optionally insert commas and periods; automatic paragraph and list structuring was not verified.	PartialSource: spokenly.app/docs — checked Oct 2026Models add punctuation, and optional Smart Paragraphs on Mac adds paragraph breaks; automatic list formatting remains unverified.	PartialSource: raw.githubusercontent.com/altic-dev — checked Oct 2026Optional AI cleanup fixes punctuation and capitalization, while spoken directives create line breaks and list items.	YesSource: docs.wisprflow.ai/articles — checked Oct 2026Smart Formatting automatically structures punctuation and lists, with controls on desktop and iOS and formatting always applied on Android.	NoSource: help.krisp.ai/hc — checked Oct 2026Not supported for system-wide dictation.
Custom dictionary	YesSource: ottex.ai/docs — checked Oct 2026Add or delete names and specialist terms with preferred spellings; entries provide recognition context during transcription.	YesSource: nuance.com/products — checked Oct 2026The Vocabulary Editor adds and trains words and phrases with custom written forms, pronunciations and formatting properties.	PartialSource: spokenly.app/docs — checked Oct 2026Supported models accept custom keyterms; ElevenLabs live transcription applies only the first 50 terms.	YesSource: fluidvoice.org/features — checked Oct 2026Custom Dictionary accepts preferred spellings for names and jargon, with JSON import and export.	YesSource: docs.wisprflow.ai/articles — checked Oct 2026Users can add names, technical terms, and jargon to an editable dictionary and configure recurring spelling replacements.	NoSource: help.krisp.ai/hc — checked Oct 2026Not supported for system-wide dictation.
Voice snippetsSay a saved phrase to insert a link or text	YesSource: ottex.ai/docs — checked Oct 2026Save a voice trigger and expansion text; saying the trigger while dictating inserts the saved URL or text in place.	YesSource: nuance.com/products — checked Oct 2026Speaking a saved Auto-Text command inserts user-defined text or graphics, with global, application-specific or window-specific availability.	NoSource: spokenly.app/comparison — checked Oct 2026Not supported.	No public claim	YesSource: docs.wisprflow.ai/articles — checked Oct 2026Speaking a saved trigger inserts its stored text during dictation; personal snippets are available on every plan.	NoSource: help.krisp.ai/hc — checked Oct 2026Not supported for system-wide dictation.
Custom voice instructionsFree-form rules applied after dictation	YesSource: ottex.ai — checked Oct 2026Saved profile instructions automatically process dictation before insertion; rules can include writing examples, personal phrasing and email signatures.	No public claim	YesSource: spokenly.app/docs — checked Oct 2026Saved mode AI Instructions process each dictation before its configured auto-insert action using the selected AI provider.	YesSource: fluidvoice.org/docs — checked Oct 2026Saved user-written prompts automatically process dictation before insertion when AI cleanup is enabled.	PartialSource: docs.wisprflow.ai/articles — checked Oct 2026Desktop beta Auto Apply After Dictation runs a saved custom prompt with optional writing samples before insertion, accepts 1–1,000 words, skips specified cleanup and rapid-repeat cases, and is subject to organization controls.	NoSource: help.krisp.ai/hc — checked Oct 2026Not supported for system-wide dictation.
Tone presetsChoose a predefined writing style	YesSelect built-in Messaging or Email profiles for casual or professional dictation; assigning apps or websites enables automatic profile activation.	No public claim	No public claim	No public claim	PartialSource: docs.wisprflow.ai/articles — checked Oct 2026Formal., Casual, very casual, and Excited! presets change capitalization and punctuation with category restrictions and English-only style notices.	NoSource: help.krisp.ai/hc — checked Oct 2026Not supported for system-wide dictation.
Text transforms / AI ShortcutsApply an AI action to existing text	YesSource: ottex.ai/docs — checked Oct 2026Select or copy existing text, open Omnibar and run an AI Shortcut to rewrite, translate, summarize or apply a custom prompt; the result is copied/pasted.	No public claim	No public claim	PartialSource: fluidvoice.org/features — checked Oct 2026Write Mode replaces selected text using spoken rewrite instructions in Accessibility-compatible fields, with AI-provider compatibility restrictions.	PartialSource: docs.wisprflow.ai/articles — checked Oct 2026Desktop beta Transforms rewrite selected or readable field text with built-in or saved prompts at 1–1,000 words; iOS Polish accepts 10–1,000 words with different controls.	PartialSource: help.krisp.ai/hc — checked Oct 2026Core and Advanced users can summarize and analyze meeting transcripts through AI Chat with saved response instructions; arbitrary selected-text actions are unverified.
Per-app output rules	YesSource: ottex.ai — checked Oct 2026Per-app output rules: exact in AI chats, polished in email.	No public claim	PartialSource: spokenly.app/docs — checked Oct 2026Mac modes automatically activate for configured applications or websites.	YesSource: fluidvoice.org/docs — checked Oct 2026User-authored per-app instructions are selected automatically for the focused application.	PartialSource: docs.wisprflow.ai/articles — checked Oct 2026Styles are selected for app categories, with web-app detection on desktop; some iOS layouts apply one default style everywhere.	NoSource: help.krisp.ai/hc — checked Oct 2026Not supported for system-wide dictation.