Ottex
Mac, iOS
You want dictated emails to follow your saved writing instructions.
You need voice dictation on a Windows computer.
Three tools. Different strengths. Find your fit.
Each tool takes a different approach. Here’s a quick side-by-side to help you choose based on your workflow, platforms, and goals.
Mac, iOS
You want dictated emails to follow your saved writing instructions.
You need voice dictation on a Windows computer.
Mac
You want free offline dictation shaped by your instructions.
You need meeting recording in a stable release.
Mac
You want unlimited meeting transcription using local models.
You need personal writing rules applied automatically during dictation.
FluidVoice is free; Hyprnote offers free on-device transcription or Pro at $140/seat/yr. Ottex offers free app access with local models or BYOK (provider charges separate), or Pro at $168/yr.
| Feature | |||
|---|---|---|---|
| Free tier | YesSource: ottex.ai — checked Oct 2026Forever free with local models and BYOK providers. | PartialSource: fluidvoice.org/download — checked Oct 2026no trial, word limit or paid tier | PartialSource: anarlog.so/pricing — checked Oct 2026Unlimited on-device transcription |
| Subscription | $14/moSource: ottex.ai — checked Oct 2026$14/mo — unlimited dictation and meeting intelligence, all hosted models included. | NoNo subscription plan | $14/moSource: anarlog.so/pricing — checked Oct 2026Published USD prices are per person for individuals and teams, with hosted transcription, AI, dictation, sync, and collaboration included. |
| One-time payment | Not applicableSource: ottex.ai — checked Oct 2026Application is free | Not applicableSource: fluidvoice.org/download — checked Oct 2026Application is free | Not applicableSource: anarlog.so/pricing — checked Oct 2026Application is free |
| Bring your own keyPay the model provider, not the middleman | YesSource: ottex.ai — checked Oct 2026OpenRouter, Google AI, OpenAI, Anthropic, Mistral, and compatible providers. | YesSource: fluidvoice.org/docs — checked Oct 2026Optional cloud cleanup uses the user's API key and separately billed provider account. | YesSource: anarlog.so/pricing — checked Oct 2026Users can supply their own transcription and Intelligence provider API keys. |
| Education discount | NoSource: ottex.ai — checked Oct 2026Same pricing for everyone. | No public claim | No public claim |
FluidVoice supports Mac with a Windows testing build; Hyprnote supports Mac with Windows and Linux betas. Ottex supports Mac and iOS, making it the choice here for iOS coverage.
| Feature | |||
|---|---|---|---|
| Platforms | Mac, iOSSource: ottex.ai — checked Oct 2026 | MacSource: fluidvoice.org/download — checked Oct 2026Some platforms in beta/waitlist | MacSource: docs.anarlog.so/desktop-installation — checked Oct 2026Some platforms in beta/waitlist |
| Native app footprintLight on memory, instant to invoke | YesSource: ottex.ai — checked Oct 2026Native Swift app, ~50MB RAM. | PartialSource: raw.githubusercontent.com/altic-dev — checked Oct 2026The macOS application uses SwiftUI and AppKit; the Windows implementation was not verified. | NoSource: github.com/fastrepl — checked Oct 2026The desktop application uses Tauri v2 with a React and TypeScript interface and Rust backend. |
FluidVoice captures audio without a bot in its Mac beta; Hyprnote does so with optional recording notices in supported Mac/Linux meeting chats. Ottex is bot-free with an opt-in, experimental Mac camera recording watermark.
| Feature | |||
|---|---|---|---|
| Meeting recordingThe call is captured at all | YesSource: ottex.ai — checked Oct 2026With speaker separation. | PartialSource: github.com/altic-dev — checked Oct 2026FluidMeet records meetings locally in macOS v1.6.10-beta.7. | YesSource: docs.anarlog.so/meetings — checked Oct 2026Meetings can be recorded from calendar events or new notes. |
| Bot-free captureNo "recording bot" joins your call | YesSource: ottex.ai — checked Oct 2026System-audio capture, no bot joins 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.anarlog.so/meetings — checked Oct 2026Desktop capture adds no participant to 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. | No public claim | No public claim |
| Participant noticesRecording notices in chat or calendar | No public claim | No public claim | PartialSource: docs.anarlog.so/automatic-capture — checked Oct 2026Users can enable recording disclosures in supported meeting chats on macOS and Linux, with no Windows support or verified calendar notice. |
| Speaker separation | YesSource: ottex.ai — checked Oct 2026 | 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.anarlog.so/models-and-providers — checked Oct 2026Supported models provide speaker diarization, and users can assign participant names to transcript labels. |
| Calendar detection | YesSource: ottex.ai — checked Oct 2026Calendar-based meeting detection. | 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.anarlog.so/calendar — checked Oct 2026Apple Calendar is supported on macOS, with read-only Google Calendar and Outlook connections. |
| Cross-meeting search | YesSource: ottex.ai — checked Oct 2026 | PartialSource: github.com/altic-dev — checked Oct 2026The macOS beta's Global Search includes transcripts across the application. | YesSource: docs.anarlog.so/chat — checked Oct 2026Chat searches meeting titles, content, and recurring meeting history. |
| Shared team memory | PartialSource: ottex.ai — checked Oct 2026Public share links, not a full team workspace. | No public claim | PartialSource: docs.anarlog.so/teams — checked Oct 2026Pro workspaces share meeting notes and new notes created in designated team folders. |
FluidVoice offers local summaries in its Mac beta. Hyprnote and Ottex structure transcripts with templates and draft follow-up emails; Ottex also has CRM drafts and reusable prompts for transcript chat.
| Feature | |||
|---|---|---|---|
| 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. | No public claim | YesSource: docs.anarlog.so/customize-summaries — checked Oct 2026The Summary picker offers named formats that structure meeting content into sections such as decisions, action items, or one-on-ones. |
| 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. | No public claim | No public claim |
| Decisions, owners, action itemsLeave the meeting with next steps clear | YesSource: ottex.ai — checked Oct 2026Decisions with owners extracted. | 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.anarlog.so/customize-summaries — checked Oct 2026Summary formats can extract action items with owners and due dates. |
| Follow-up email draftedThe outcome is sent, not summarized | YesSource: ottex.ai — checked Oct 2026Drafted from templates or your own. | No public claim | YesSource: docs.anarlog.so/chat — checked Oct 2026Chat can draft a follow-up email from the open meeting. |
| CRM update / ticket draft | YesSource: ottex.ai — checked Oct 2026Structured for the target app. | No public claim | PartialSource: docs.anarlog.so/automations — checked Oct 2026Pro automations create Linear issue drafts from assigned meeting follow-ups. |
| Your rough notes steer the result | YesSource: ottex.ai — checked Oct 2026Rough notes and instructions steer the result. | No public claim | YesSource: docs.anarlog.so/meetings — checked Oct 2026User-written Memos provide context for generated meeting summaries. |
FluidVoice and Ottex automatically apply saved instructions to dictation and use custom dictionaries. Ottex formats dictation into lists and inserts saved text via voice triggers; Hyprnote has Pro dictation, but these controls are unverified.
| Feature | |||
|---|---|---|---|
| Engines | Cloud: Whisper, Gemini 3 Flash, Voxtral. Local: Whisper, Parakeet v3.Source: ottex.ai — checked Oct 2026Hosted routing, BYOK, or fully local. | 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. | Configurable hosted or local transcription providersSource: docs.anarlog.so/dictation — checked Oct 2026Dictation uses the selected Transcription provider. |
| Local / on-device optionAudio never leaves the machine | YesSource: ottex.ai — checked Oct 2026Whisper and Parakeet v3 run fully on-device. | YesSource: fluidvoice.org — checked Oct 2026Local recognition and local cleanup work offline after their required assets are installed. | PartialSource: docs.anarlog.so/dictation — checked Oct 2026A supported local transcription provider keeps dictation processing on the computer. |
| Model selection per task | YesSource: ottex.ai — checked Oct 2026Choose the model per task and per provider. | YesSource: fluidvoice.org/models — checked Oct 2026Users can download and switch speech models in the application. | PartialSource: docs.anarlog.so/dictation — checked Oct 2026The dictation setup explicitly allows choosing a transcription provider. |
| Languages & mixed speechWork that crosses languages mid-sentence | YesSource: ottex.ai — checked Oct 2026100+ languages · Mid-sentence language switching. | PartialSource: fluidvoice.org/docs — checked Oct 202699+ languages · Mid-sentence language switching can produce errors and requires testing with the selected model. | No public claim |
| 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: raw.githubusercontent.com/altic-dev — checked Oct 2026Optional AI cleanup fixes punctuation and capitalization, while spoken directives create line breaks and list items. | No public claim |
| 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: fluidvoice.org/features — checked Oct 2026Custom Dictionary accepts preferred spellings for names and jargon, with JSON import and export. | No public claim |
| 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. | No public claim | No public claim |
| 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. | YesSource: fluidvoice.org/docs — checked Oct 2026Saved user-written prompts automatically process dictation before insertion when AI cleanup is enabled. | No public claim |
| 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 |
| 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. | 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.anarlog.so/chat — checked Oct 2026Meeting Chat can propose summary edits and transcript corrections for user review. |
| Per-app output rules | YesSource: ottex.ai — checked Oct 2026Per-app output rules: exact in AI chats, polished in email. | YesSource: fluidvoice.org/docs — checked Oct 2026User-authored per-app instructions are selected automatically for the focused application. | No public claim |
Choose FluidVoice for free offline dictation with saved writing rules; choose Hyprnote for meeting transcription, structured summaries and follow-up drafts. Choose Ottex for automatic dictation rules, voice-triggered saved text and meeting-derived CRM drafts.
Yes: FluidVoice is free without a trial or word cap, and Hyprnote includes unlimited on-device transcription, while system-wide dictation requires Pro. Ottex has free local-model and BYOK app access; BYOK provider charges are separate.
FluidVoice costs $0 and has no paid tier; Hyprnote Pro costs $14/seat/mo or $140/seat/yr, with free on-device transcription. Ottex Pro costs $14/mo or $168/yr, with free local-model or BYOK app access and separate provider charges for BYOK.
For audio privacy, FluidVoice can run recognition and cleanup locally; Hyprnote offers on-device transcription, and Ottex runs Whisper or Parakeet on-device. All three support BYOK, so select a local configuration when you want audio processed on your computer.
Hyprnote offers meeting templates and follow-up email drafts, so switch if those matter more than saved dictation rules. Stay with FluidVoice for free offline dictation shaped by your instructions, or choose Ottex for writing rules plus meeting-derived CRM and ticket drafts.
Both have Windows prereleases: FluidVoice has a 64-bit testing build, and Hyprnote has a Windows beta plus a Linux beta. Ottex supports Mac and iOS without Windows support; Android support is unverified for all three.
Ottex lists 100 languages and supports mid-sentence switching; FluidVoice lists 99, but switching can produce model-dependent errors. Hyprnote has no verified language count or mid-sentence switching support.