Open-source voice dictation recommendations from the X thread

A detailed pass over the replies to Parth Jadhav's question about open-source alternatives to Wispr Flow, with source links, repo evidence, and a feature-by-feature comparison matrix.

97X records captured after expanding replies and probable spam
29products and reference options compared
23public repositories inspected or cloned
24features and capabilities checked

Method

I captured the X thread in Chrome, expanded hidden reply pockets and probable-spam replies, resolved the shortened links, cloned the public repositories I could identify, read the READMEs, and searched code where the README did not expose enough feature detail. The table below uses Yes, No, Partial, Planned, and Unknown; strict yes/no would overstate several closed-source or sparsely documented products.

Notable code dives: Handy for post-processing providers, custom words, history, and translation settings; Freestyle for local Whisper, local LLM, dictionary, context, and history; Amical for paste helpers, shortcuts, selection context, and planned MCP/meeting features.

All recommendations and source links

RecommendationLinksRepo metadataX comments
HandyOpen-source cross-platform dictation site/docs
repository		 22,814 stars · MIT License · latest v0.8.3 (2026-04-28) source 1 source 2 source 3 source 4 source 5 source 6 source 7 source 8 source 9 source 10 source 11
VoiceInkNative macOS dictation site/docs
repository		 5,141 stars · Other · latest v1.79 (2026-05-23) source 1 source 2 source 3 source 4 source 5 source 6
MuesliLocal-first macOS dictation and meetings site/docs
repository		 537 stars · MIT License · latest v0.6.9 (2026-05-27) source 1 source 2 source 3 source 4
OpenWhisprCross-platform dictation and AI notes site/docs
repository		 3,472 stars · MIT License · latest v1.7.2 (2026-05-20) source 1 source 2 source 3 source 4 source 5 source 6
thinkurmacOS dictation and meeting transcripts site/docs
repository		 19 stars · MIT License source 1
Freestyle VoiceSource-available local-first dictation site/docs
repository		 126 stars · MIT License · latest 0.0.10 (2026-05-31) source 1 source 2 source 3
AmicalLocal-first dictation and note-taking site/docs
repository		 1,292 stars · MIT License · latest v1.7.1 (2026-05-26) source 1
MacParakeetApple Silicon Parakeet dictation and transcription site/docs
repository		 285 stars · Other · latest v0.6.16 (2026-05-31) source 1
FluidVoicemacOS dictation with command mode site/docs
repository		 2,269 stars · GNU General Public License v3.0 · latest v1.5.14 (2026-05-29) source 1
OpenLessmacOS and Windows voice input site/docs
repository		 1,931 stars · MIT License · latest v1.3.4-tauri (2026-05-22) source 1
TypefluxmacOS local-first dictation and Ask Anything site/docs
repository		 271 stars · GNU Affero General Public License v3.0 source 1
type4memacOS voice input with many providers site/docs
repository		 1,226 stars · MIT License · latest v1.9.3 (2026-04-26) source 1
input0macOS voice input site/docs
repository		 266 stars · Other · latest v0.6.1 (2026-05-28) source 1
Speed of SoundLinux desktop voice typing site/docs
repository		 144 stars · MIT License · latest v0.14.0 (2026-05-09) source 1
MoFA-IMEmacOS local voice input method site/docs
repository		 4 stars · Apache License 2.0 · latest 0.3.0 (2026-02-23) source 1
VOICE2TYPEWindows tray voice input site/docs
repository		 38 stars · MIT License · latest v0.0.51-the-last-pure-vision (2026-05-26) source 1
WhispmacOS local Kyutai dictation site/docs
repository		 0 stars source 1
whispyServerless Groq dictation site/docs
repository		 2 stars source 1
InvokeVoice command layer site/docs
repository		 3 stars source 1
TMSpeechWindows real-time subtitles site/docs
repository		 1,332 stars · MIT License · latest v0.4.3 (2025-12-15) source 1
BreezeAppMobile offline AI app site/docs
repository		 221 stars · Apache License 2.0 · latest v2.0.0 (2025-08-05) source 1
Speech-to-clipboardLinux speech-to-clipboard script site/docs
repository		 0 stars · MIT License source 1
ShuoClosed app, public website repo site/docs
repository		 0 stars · Apache License 2.0 source 1
WalkieClosed-source desktop speech-to-text site/docs No public repo metadata source 1
VoibeClosed-source macOS voice input site/docs No public repo metadata source 1
SuperwhisperClosed-source macOS/iOS dictation site/docs No public repo metadata source 1
TypelessClosed-source voice typing product site/docs No public repo metadata source 1 source 2
Raycast DictationFeature inside Raycast site/docs No public repo metadata source 1
macOS Voice ControlBuilt-in OS accessibility feature site/docs No public repo metadata source 1

Feature and capability matrix

Yes No Partial Planned Unknown
Feature / capabilityHandyVoiceInkMuesliOpenWhisprthinkurFreestyle VoiceAmicalMacParakeetFluidVoiceOpenLessTypefluxtype4meinput0Speed of SoundMoFA-IMEVOICE2TYPEWhispwhispyInvokeTMSpeechBreezeAppSpeech-to-clipboardShuoWalkieVoibeSuperwhisperTypelessRaycast DictationmacOS Voice Control
Open or public sourcePublic repository and inspectable source. Partial means source-available or unclear license. YesYesYesYesYesYesYesYesYesYesYesYesPartialYesYesYesYesYesYesYesYesYesNoNoNoNoNoNoNo
macOS appRuns as a desktop dictation or voice app on macOS. YesYesYesYesYesYesYesYesYesYesYesYesYesNoYesNoYesYesPartialNoNoNoYesYesYesYesYesYesYes
Windows appRuns on Windows as a desktop app or tray utility. YesNoNoYesNoYesNoNoNoYesNoNoNoNoNoYesNoYesYesYesNoNoYesYesNoNoYesNoNo
Linux appRuns on Linux as a desktop app or daemon. YesNoNoYesNoYesNoNoNoPartialNoNoNoYesNoNoNoNoNoNoNoYesNoNoNoNoNoNoNo
Mobile appAndroid or iOS support was documented. NoNoNoPartialNoNoPartialNoNoNoNoNoNoNoNoNoNoNoYesNoYesNoNoNoNoYesYesNoNo
System-wide dictationCan enter dictated text into arbitrary apps, not only inside its own transcript view. YesYesYesYesYesYesYesYesYesYesYesYesYesYesYesYesYesYesPartialNoNoPartialYesYesYesYesYesPartialYes
Global hotkey or push-to-talkGlobal shortcut, hold-to-talk, toggle recording, or equivalent. YesYesYesYesYesYesYesYesYesYesYesYesYesYesYesYesYesYesYesUnknownNoPartialUnknownYesYesYesYesYesPartial
Auto paste or direct insertionAutomatically inserts text at cursor or uses clipboard/keyboard injection. YesYesYesYesYesYesYesYesYesYesYesYesYesYesYesYesYesYesPartialNoNoPartialYesYesYesYesYesPartialYes
Local/offline STT pathSpeech-to-text can run fully on device. YesYesYesYesYesYesYesYesYesYesYesYesYesYesYesYesYesNoYesYesYesNoYesYesYesYesUnknownUnknownPartial
Cloud or BYOK STT pathSpeech-to-text can use hosted APIs or user-supplied API keys. NoNoPartialYesNoYesNoNoNoYesYesYesPartialNoNoYesNoYesNoNoNoYesPartialYesUnknownPartialUnknownUnknownUnknown
Whisper supportOpenAI Whisper, whisper.cpp, WhisperKit, Foundry Whisper, or compatible models. YesYesYesYesNoYesYesPartialYesYesYesNoYesYesYesYesNoYesYesNoNoNoUnknownYesUnknownYesUnknownUnknownNo
Parakeet supportNVIDIA Parakeet or Parakeet TDT/V3/V2 support. YesYesYesYesUnknownNoNoYesYesNoNoNoNoYesNoNoNoNoNoNoNoNoUnknownYesUnknownUnknownUnknownUnknownNo
Other local STT modelsExamples include Apple Speech, Kyutai, sherpa-onnx, SenseVoice, FunASR, Qwen3-ASR, Cohere local paths. PartialPartialYesYesYesPartialPartialPartialYesYesYesYesYesYesPartialNoYesNoPartialYesYesNoUnknownYesUnknownPartialUnknownUnknownYes
AI cleanup and formattingPost-processing, rewrite modes, filler removal, grammar fixes, or style templates. YesYesYesYesYesYesYesYesYesYesYesYesYesYesYesPartialNoYesYesPartialPartialPartialYesYesYesYesYesPartialNo
Custom vocabulary or dictionaryHotwords, replacements, snippets, personal vocabulary, or custom words. YesYesYesUnknownPartialYesUnknownYesUnknownYesPartialYesYesYesUnknownNoNoNoYesNoNoNoUnknownYesUnknownPartialUnknownUnknownPartial
Context-aware outputUses active app, selected text, screen content, URL, per-app styles, or task context. NoYesPartialPartialYesYesYesPartialNoPartialYesPartialNoPartialNoNoNoNoPartialNoNoNoNoPartialPartialPartialPartialYesNo
Transcript history or libraryStores searchable history, notes, previous transcripts, or session records. YesUnknownYesYesYesYesYesYesYesYesYesYesYesUnknownYesPartialNoUnknownUnknownYesUnknownNoUnknownYesUnknownYesUnknownPartialNo
Meeting recording/transcriptionCan record meetings, system audio, or live meeting sessions. NoNoYesYesYesNoPlannedYesNoNoNoNoNoNoNoPartialNoNoNoYesNoNoNoNoNoUnknownNoNoNo
Speaker labels or diarizationSpeaker separation, labels, diarization, or meeting-speaker support. NoNoYesYesYesNoNoPartialNoNoNoNoNoNoNoNoNoNoNoNoNoNoNoNoNoUnknownNoNoNo
File/audio/video importTranscribes existing audio/video files, URLs, or YouTube links. NoNoNoYesNoNoNoYesNoNoNoNoNoNoNoNoNoNoNoNoPartialNoNoNoUnknownYesUnknownUnknownNo
Translation modeBuilt-in translate-to-English or translation workflow. PartialNoNoUnknownNoNoNoNoNoYesYesYesNoNoNoYesNoNoNoNoNoNoNoYesUnknownPartialUnknownPartialNo
Voice commands or agentsCommands, app actions, command modes, voice assistant, or agent workflows. NoPartialNoYesNoNoYesPartialYesPartialYesPartialNoNoNoNoNoNoYesNoPartialNoNoYesPartialPartialPartialYesYes
Automation API, CLI, MCP, or pluginsCLI, MCP, API, Raycast integration, Composio, or similar automation hooks. YesNoYesYesYesUnknownPlannedYesNoNoNoPartialNoNoNoNoNoNoYesNoNoPartialNoYesNoPartialNoYesYes
Private local-first pathA documented path with local processing and no required account/cloud for core use. YesYesYesYesYesPartialYesYesPartialPartialPartialPartialPartialYesYesPartialYesNoPartialYesYesNoPartialPartialPartialPartialUnknownUnknownPartial

Detailed product notes

Primary OSS recommendations

Handy

Open-source cross-platform dictation

site/docsrepo

The most repeated recommendation. README positions it as a free offline speech-to-text app for macOS, Windows, and Linux with auto-paste, configurable shortcuts, Whisper and Parakeet models. Code search also showed post-processing providers, custom words, history, translation settings, Raycast integration, and CLI flags.

22,814 stars · MIT License · latest v0.8.3 (2026-04-28)

Evidence and source comments
  • README: offline operation, auto-paste, global shortcut, Whisper small/medium/turbo/large, Parakeet V3, custom Whisper models, VAD, Raycast, CLI.
  • Code: post-process providers include OpenAI, Groq, and custom providers; custom words, history entries, and translation settings are present.

source 1 source 2 source 3 source 4 source 5 source 6 source 7 source 8 source 9 source 10 source 11

VoiceInk

Native macOS dictation

site/docsrepo

Native macOS voice-to-text focused on privacy and local AI. It is strong on local transcription, smart modes, personal dictionary, contextual formatting, and screen/app awareness, but is macOS-only.

5,141 stars · Other · latest v1.79 (2026-05-23)

Evidence and source comments
  • README: 100 percent offline, local AI models, Power Mode app/URL detection, screen-content context, configurable shortcuts, personal dictionary, smart modes, and AI assistant.
  • Technology list includes whisper.cpp and FluidAudio/Parakeet.

source 1 source 2 source 3 source 4 source 5 source 6

Muesli

Local-first macOS dictation and meetings

site/docsrepo

A rich local-first macOS option that goes beyond dictation into meeting capture, diarization, note generation, calendar context, exports, and model management.

537 stars · MIT License · latest v0.6.9 (2026-05-27)

Evidence and source comments
  • README: on-device speech-to-text, mic plus system audio meeting recording, VAD chunking, speaker diarization, meeting detection, Google Calendar integration, PDF/Markdown export, templates, filler removal, personal dictionary, configurable hotkeys, CLI.
  • Models/providers: Parakeet TDT, Cohere Transcribe, WhisperKit, Qwen3; AI notes via BYOK OpenAI/OpenRouter/ChatGPT OAuth/local Ollama.

source 1 source 2 source 3 source 4

OpenWhispr

Cross-platform dictation and AI notes

site/docsrepo

Cross-platform open-source app aimed at voice-to-text, meetings, AI notes, semantic search, and local/cloud model choice. One of the more feature-complete options on paper.

3,472 stars · MIT License · latest v1.7.2 (2026-05-20)

Evidence and source comments
  • README/site: macOS, Windows, Linux; local or cloud; Whisper and Parakeet; speaker diarization; meeting transcription; Google Calendar; notes, semantic search, API/MCP.
  • Public releases exist for multiple platforms.

source 1 source 2 source 3 source 4 source 5 source 6

thinkur

macOS dictation and meeting transcripts

site/docsrepo

macOS-only, local-first dictation and meeting recorder. The source comment mentioned local models, but the current README/site primarily describe Apple's Speech framework, meeting transcripts, speaker labels, search, MCP, and per-app styles.

19 stars · MIT License

Evidence and source comments
  • README/site: free open-source macOS voice-to-text, dictation plus meeting recording, 100 percent local, Apple Speech framework, speaker labels, searchable transcripts, MCP, smart post-processing, self-correction, per-app styles.
  • Constraint: macOS 15+ and Apple Silicon.

source 1

Freestyle Voice

Source-available local-first dictation

site/docsrepo

Local-first dictation for macOS, Windows, and Linux. README emphasizes cloud-provider choice, cleaning, dictionary, contextual correction, and cross-platform releases; code search showed local Whisper support and local LLM config.

126 stars · MIT License · latest 0.0.10 (2026-05-31)

Evidence and source comments
  • README: hotkey, paste at cursor, providers OpenAI/Groq/Anthropic/Google/Deepgram/ElevenLabs, transcription cleaning, dictionary, contextual correction, macOS/Windows/Linux releases.
  • Code search: whisper.cpp download/use, history, dictionary, context-aware dictation, local LLM config.

source 1 source 2 source 3

Amical

Local-first dictation and note-taking

site/docsrepo

Open-source AI dictation and note-taking with context-aware dictation, active-app formatting, hotkeys, local models, and a native helper. Meeting transcription and MCP are marked as planned in its own feature table.

1,292 stars · MIT License · latest v1.7.1 (2026-05-26)

Evidence and source comments
  • README: local-first, Whisper/Ollama, context-aware dictation, active app formatting, extensible hotkeys/voice macros/custom workflows, local/offline models, floating widget.
  • README feature table marks MCP and meeting transcription as planned/partial rather than done. Code search showed paste helper, shortcuts, audio mute/restore, and selected-text context.

source 1

MacParakeet

Apple Silicon Parakeet dictation and transcription

site/docsrepo

Mac app centered on Parakeet running locally via FluidAudio/ANE. It is unusually broad for transcription input: system dictation, file/URL/YouTube transcription, meeting recording, calendar reminders, text cleanup, replacements, snippets, summaries, chat, and CLI.

285 stars · Other · latest v0.6.16 (2026-05-31)

Evidence and source comments
  • README: Parakeet TDT on ANE via FluidAudio, system-wide dictation, file/URL/YouTube transcription, meeting recording, calendar reminders/autostart, text cleanup, replacements/snippets, summaries/chat/formatter/transforms.
  • Also supports optional WhisperKit; Apple Silicon and macOS 14.2+.

source 1

FluidVoice

macOS dictation with command mode

site/docsrepo

Open-source macOS voice-to-text with strong local model support and optional AI enhancement. It has live preview, smart typing into any app, command mode, write mode, and history/stats.

2,269 stars · GNU General Public License v3.0 · latest v1.5.14 (2026-05-29)

Evidence and source comments
  • README: Parakeet Flash/v3/v2, Cohere Transcribe, Apple Speech, Whisper; live preview; low latency; OpenAI/Groq/custom providers for enhancement; global hotkey; smart typing into any app; menu bar; command mode; write mode.
  • macOS 15+, Apple Silicon preferred; Intel supported with Whisper models in later versions.

source 1

OpenLess

macOS and Windows voice input

site/docsrepo

Open-source voice input for macOS and Windows with hotkey dictation, cursor insertion, AI polishing modes, translation, selection QA, history, vocabulary, and both local and cloud ASR paths.

1,931 stars · MIT License · latest v1.3.4-tauri (2026-05-22)

Evidence and source comments
  • README: macOS and Windows; hotkey; AI-polished text; insert at cursor; clipboard fallback; cloud ASR via Volcengine/OpenAI Whisper-compatible/Apple Speech; local ASR via Qwen3-ASR and Windows Foundry Local Whisper.
  • README: polish providers Ark/DeepSeek/OpenAI/Doubao/Anthropic-compatible/custom; raw/light/structured/formal modes; translation hotkey; selection-ask QA panel; history, vocab, settings, hotwords.

source 1

Typeflux

macOS local-first dictation and Ask Anything

site/docsrepo

macOS voice input with hold-Fn dictation and double-press Ask Anything. It combines local STT options with many cloud providers, custom personas, streaming preview, history, translation, and export.

271 stars · GNU Affero General Public License v3.0

Evidence and source comments
  • README: hold Fn to dictate, double press for Ask Anything, text into any macOS app, local-first/privacy-first, custom personas, multiple speech backends.
  • Local models include SenseVoice, FunASR, WhisperKit, Qwen3-ASR; cloud options include Alibaba, Doubao, Google, OpenAI, Groq, and more.

source 1

type4me

macOS voice input with many providers

site/docsrepo

macOS voice input with built-in local recognition, cloud providers, streaming recognition, prompt templates, hotwords/snippets, local LLM options, and history export.

1,226 stars · MIT License · latest v1.9.3 (2026-04-26)

Evidence and source comments
  • README: built-in local recognition plus cloud providers; streaming recognition; text polish, prompt optimization, translation, custom templates; Ollama local LLM; hotwords/snippet replacements; history export.
  • Local version mentions SenseVoice and Qwen3-ASR; architecture lists 13 LLM providers.

source 1

input0

macOS voice input

site/docsrepo

macOS voice input: hold a hotkey, transcribe locally, refine with LLM, and auto-paste. License metadata is 'Other', so treat it as public-source/source-available rather than cleanly OSI-open.

266 stars · Other · latest v0.6.1 (2026-05-28)

Evidence and source comments
  • GitHub metadata description: hold a hotkey to record, release to transcribe locally via STT, refine with LLM, and auto-paste into the active text field.
  • README/repo includes app flow around local STT, LLM refinement, history, custom words, and macOS integration.

source 1

Speed of Sound

Linux desktop voice typing

site/docsrepo

A Linux-first voice typing app with offline local models, direct typing into apps, tray/global shortcut support, optional LLM polishing, and custom context/vocabulary.

144 stars · MIT License · latest v0.14.0 (2026-05-09)

Evidence and source comments
  • README: Linux desktop voice typing, offline transcription via Whisper/Parakeet/Canary and other models, global shortcut/system tray/button, X11/Wayland portals, multi-language.
  • Optional LLM polishing via Anthropic/Google/OpenAI or self-hosted vLLM/Ollama/llama.cpp; supports custom context/vocab.

source 1

MoFA-IME

macOS local voice input method

site/docsrepo

macOS resident voice input: hold Fn, locally transcribe, optionally polish with local LLM, and write into the current input field. It also includes model management and history/clipboard utilities.

4 stars · Apache License 2.0 · latest 0.3.0 (2026-02-23)

Evidence and source comments
  • README: hold fn to local transcribe, optional local LLM polish, writes into input field, global hotkey through CGEventTap, menu bar, floating UI, history window.
  • Local chain: Whisper ASR and Qwen GGUF through llama.cpp FFI; model management GUI and clipboard manager.

source 1

VOICE2TYPE

Windows tray voice input

site/docsrepo

Windows tray voice input with hotkey/toggle modes, clipboard paste or keyboard injection, Groq/SiliconFlow/local Whisper engines, and experimental real-time subtitles.

38 stars · MIT License · latest v0.0.51-the-last-pure-vision (2026-05-26)

Evidence and source comments
  • README: global hotkey, hold-to-talk/toggle, clipboard paste/keyboard injection, SiliconFlow/Groq/local Whisper, filtering punctuation/emoji, re-paste last, autostart/update/logs.
  • Experimental real-time subtitles from speaker/mic audio and bilingual subtitles through LLM providers.

source 1

Whisp

macOS local Kyutai dictation

site/docsrepo

Small macOS push-to-talk app that runs locally on Apple Silicon using Kyutai STT via MLX. It is privacy-focused but intentionally minimal compared with the feature-heavy alternatives.

0 stars

Evidence and source comments
  • README: hold Fn, locally on Apple Silicon using Kyutai STT through MLX, menu bar, no cloud, macOS 13+, Apple Silicon, roughly 3 GB model.

source 1

whispy

Serverless Groq dictation

site/docsrepo

A simple serverless dictation app for macOS and Windows using Groq Whisper Large v3, optional OpenAI cleanup, hotkeys, auto-paste, and tray mode. The author explicitly said it does not support local models.

2 stars

Evidence and source comments
  • README: Groq Whisper Large v3, optional OpenAI cleanup, hotkeys, auto-paste, tray, Windows/macOS.
  • Thread reply from author: it does not support local models.

source 1

Invoke

Voice command layer

site/docsrepo

More of a local-first voice command/action layer than a Wispr Flow clone. It maps speech to structured actions across apps using local STT/LLM plus Composio integrations.

3 stars

Evidence and source comments
  • README: desktop/Windows/Android voice command layer, natural speech to structured actions, local Whisper tiny STT, Qwen 3 0.6B through Ollama, Composio integrations for Gmail/GitHub/Slack/Calendar/Notion/Todoist/Docs/search.
  • Also includes writing cleanup, snippets, dictionary, style presets, privacy mode, Android voice bubble, Windows Tauri desktop.

source 1

TMSpeech

Windows real-time subtitles

site/docsrepo

Windows real-time Chinese subtitle and meeting transcription tool. It is adjacent to dictation rather than a cursor-insertion replacement, but relevant for local meeting speech capture.

1,332 stars · MIT License · latest v0.4.3 (2025-12-15)

Evidence and source comments
  • README: Windows realtime Chinese subtitles from system audio via WASAPI loopback, lyric-style subtitles, meeting live transcription, meeting minutes saved to file, sherpa-onnx streaming model.

source 1

BreezeApp

Mobile offline AI app

site/docsrepo

Mobile offline AI platform for Android/iOS with speech-to-text among other modalities. Relevant to voice transcription, but not a desktop system-wide dictation app.

221 stars · Apache License 2.0 · latest v2.0.0 (2025-08-05)

Evidence and source comments
  • README: Android/iOS, offline speech-to-text, text-to-speech, chat, image QA, privacy/offline mobile AI platform.

source 1

Speech-to-clipboard

Linux speech-to-clipboard script

site/docsrepo

Small Linux script that records audio, sends it to ElevenLabs STT, cleans incidental text, and writes the result to the clipboard. Useful but much narrower than a full app.

0 stars · MIT License

Evidence and source comments
  • README: Linux script, speech to clipboard, ElevenLabs API, can be bound to a hotkey, includes lightweight cleanup; local model support would require modification.

source 1

Shuo

Closed app, public website repo

site/docsrepo

The tweet listed Shuo, but also said the app is work-in-progress and will not be open source in the near future. The cloned repo is only the website, so product claims are less inspectable.

0 stars · Apache License 2.0

Evidence and source comments
  • Website README: professional local voice input method for macOS/Windows, local voice recognition, custom AI providers, auto-structuring, colloquial filtering, privacy.
  • Thread caveat: app itself is not expected to be open source soon.

source 1

Closed/reference suggestions

Walkie

Closed-source desktop speech-to-text

site/docs

Closed-source macOS/Windows desktop STT suggested in the thread. Official site describes fast cloud mode and local mode, multiple STT models, grammar cleanup, dictionary, snippets, task switching, translation, MCP connections, assistant mode, and history.

Evidence and source comments
  • Official page metadata: desktop speech-to-text with Fast Mode for cloud transcription and formatting, and Local Mode for fully on-device dictation; macOS and Windows.
  • Official embedded site copy referenced Whisper, Parakeet, Moonshine, SenseVoice, Breeze ASR, grammar fixing, filler removal, custom dictionary, snippets, translation, BYOK, MCP, and history.

source 1

Voibe

Closed-source macOS voice input

site/docs

Closed-source macOS option mentioned in the thread. Public materials describe local/offline private dictation with AI text enhancement and developer-oriented workflows, but source could not be inspected.

Evidence and source comments
  • Official site and comparison references describe a private/offline macOS dictation product with AI polish and developer modes.
  • No public repository found.

source 1

Superwhisper

Closed-source macOS/iOS dictation

site/docs

Closed-source reference product. Useful benchmark for macOS/iOS dictation, local transcription, custom prompts, and app workflows, but not an open-source alternative.

Evidence and source comments
  • Official docs page fetched for product behavior; thread mentioned it as a non-OSS option.
  • No public source code found.

source 1

Typeless

Closed-source voice typing product

site/docs

Mentioned as a reference or requested open-source-like option. It is not public source in the material found, so it is included as a benchmark rather than a candidate OSS replacement.

Evidence and source comments
  • Official pages were fetched, but no public repository was found.
  • Thread comments framed it as a product to use or emulate, not as an open-source project.

source 1 source 2

Raycast Dictation

Feature inside Raycast

site/docs

Suggested as an existing workflow inside Raycast rather than an open-source standalone replacement. Best treated as a benchmark for quick macOS productivity integration.

Evidence and source comments
  • Thread mention was direct but no open-source repository was found for the dictation feature.
  • Raycast AI is a macOS productivity product with voice/dictation-style workflows.

source 1

macOS Voice Control

Built-in OS accessibility feature

site/docs

Built-in accessibility/dictation option suggested as a baseline. It is not open-source and lacks modern AI cleanup, but it gives macOS users a no-install local-ish alternative.

Evidence and source comments
  • Apple support documentation describes Voice Control for macOS.
  • Thread mention framed it as the built-in alternative to try.

source 1

Additional mentions not fully comparable

NVIDIA Parakeet v3

Model/building block

Frequently mentioned as the local STT model to build on. It is not itself a full dictation app, but it appears in Handy, Muesli, MacParakeet, FluidVoice, OpenWhispr, Speed of Sound, and Walkie claims.

source 1 source 2 source 3 source 4

Dictify

Unresolved public repo

The quote described an open-source macOS voice-to-text app, but the shortened link did not resolve to a public repository during this pass.

source 1

Maven

Unresolved product/repo

Mentioned as a free open-source agent harness with native voice support. I could not identify a definitive public repo from the available thread context.

source 1

Codex Desktop STT

Feature mention

Mentioned as a useful in-app speech-to-text experience, but I did not find public docs positioning it as a standalone/system-wide Wispr Flow replacement.

source 1 source 2

Cloudflare Workers DIY

DIY idea

A suggestion to build a serverless DIY flow rather than use a packaged product.

source 1

Private Groq script / 'groq whisker'

Private implementation

Mentioned as a private script/repo, so it is not comparable from public artifacts.

source 1