Open source alternative to Wispr Flow, Superwhisper, Willow Voice and Dragon Professional
Last verified
Run push-to-talk voice typing on Windows with no account or subscription, transcribing into any focused app via cloud, Whisper, or fully offline.
PipeVoice was built to voice-dictate into Claude Code, Cursor, and similar terminals. Set a per-app profile for the terminal (raw text plus auto-Enter) so spoken prompts land and submit without touching the keyboard. Flow mode cleans up prompt phrasing before it reaches the agent.
You need a cross-platform tool that works on macOS or Linux. PipeVoice is Windows-only.
Users paying for Wispr Flow or Dragon Professional on Windows can switch to PipeVoice and cover the same daily dictation use case with no recurring cost. Cloud transcription through Deepgram or OpenAI costs pennies a day at conversational volumes; local Whisper removes even that.
You need a macOS tool or a managed service with a support team and guaranteed uptime. PipeVoice is self-run with no commercial support.
The meeting feature records both sides of a call as separate audio streams, transcribes them with speaker names you assign, and generates bullet points or action items with owners. Bookmarks let you flag key moments during the call by saying 'bookmark that'; the summary prioritizes flagged moments. Recordings stay on your machine unless you choose a cloud engine.
You need automatic speaker diarization across a large team or integration with a note-taking platform. PipeVoice's meeting feature is a local transcript and summary tool, not a team collaboration layer.
Voice typing on Windows has meant choosing between expensive subscriptions, bulky local installs, or apps that require a GPU and gigabyte-scale model downloads before anything works. Paid services like Wispr Flow charge monthly fees; Dragon Professional requires a large upfront license and a proprietary acoustic model that can struggle with non-standard accents.
The gap is sharpest for developers who want to dictate prompts into AI coding agents like Claude Code or Cursor. That workflow requires injecting keystrokes into a terminal, running without a mandatory account, and optionally operating fully offline. No major Windows dictation app covered that combination cleanly.
Choose Deepgram for live word-by-word streaming (fastest, cloud-based, ~$0.0043/min), OpenAI Whisper for high accuracy (~$0.006/min), or a local Whisper model on CPU for 100% offline operation. All three are selectable from the tray without restarting the app. If a cloud engine fails due to no internet, PipeVoice retries the utterance on local Whisper automatically.
Assign a different engine, cleanup setting, output mode, and auto-Enter behavior to each application. A terminal profile can receive raw text with an automatic Enter press; a chat app can receive polished, auto-sent text; a code editor can skip AI cleanup entirely. Profiles match by the focused window's application name.
Records your microphone and your computer's audio output as separate streams, so your voice and other speakers are always distinct without diarization guesswork. After recording, transcribe with Deepgram (remote speaker separation) or local Whisper (fully offline), name speakers by clicking their label, then generate bullet points, to-dos, or action items with named owners.
After transcription, an optional cleanup step removes filler words and fixes punctuation using OpenAI, Google Gemini's free tier, OpenRouter free models, or offline Ollama. Speech notes let you describe your accent or stutter so the cleanup model handles your speech accurately. Cleanup runs per-app and can be disabled for terminals or editors where raw output is preferred.
Enable Agent MCP in the tray to expose two local tools to any MCP-compatible agent: `listen` lets the agent ask a question you answer by voice, and `transcribe` converts a local audio or video file to text with word-level timestamps, SRT, or VTT captions. No API key is required for either tool; both run on the local machine.
Wispr Flow and PipeVoice both transcribe voice into any focused window, but they serve different deployment models. Wispr Flow is a paid subscription SaaS available on macOS and Windows; PipeVoice is free, MIT-licensed, and Windows-only.
| Feature | PipeVoice | Wispr Flow |
|---|---|---|
| Price | Free (MIT) | Paid subscription |
| Platform | Windows 10 and 11 | macOS and Windows |
| Account required | No | Yes |
| Offline mode | Yes (local Whisper) | No |
| Engine choice | Cloud or local, switchable | Cloud only |
| AI cleanup | Gemini free, OpenAI, Ollama | Built-in |
PipeVoice is the better choice for Windows users who want no subscription, an offline fallback, or the ability to switch transcription engines at runtime. Wispr Flow is worth considering when you need macOS support or a fully managed service with no configuration.
Superwhisper is a macOS voice typing tool built around Whisper. The comparison is primarily a platform question: Superwhisper runs on macOS; PipeVoice runs on Windows. Both offer local Whisper transcription. PipeVoice adds Deepgram live streaming and an online AI cleanup layer. If you are on macOS, Superwhisper is the natural choice. For Windows, PipeVoice covers the same use case without a subscription.
Dragon Professional is a long-established proprietary dictation product for Windows with a large one-time license cost and an acoustic model optimized for professional transcription. PipeVoice covers everyday push-to-talk dictation and AI coding workflows for free, with no acoustic model to maintain. Dragon Professional is worth considering if you need enterprise-grade medical or legal transcription accuracy or deep Microsoft Office integration. For general dictation into terminals and chat apps, PipeVoice covers the use case at no cost.
Willow Voice is a paid voice typing tool for Windows. PipeVoice covers the same push-to-talk-into-any-window use case with an MIT license, no subscription, and additional options for local transcription and AI coding workflows. Willow Voice may be worth considering if you prefer a commercial product with dedicated support. For users comfortable running an open source Windows app, PipeVoice offers the same core capability for free.
Download the Windows installer from GitHub Releases: https://github.com/Powleads/PipeVoice/releases/latestYes. Choose the Local Whisper engine and optionally Ollama for AI cleanup, and nothing leaves your machine. No internet connection and no API key are required. The local Whisper model is ~150 MB and runs on CPU with no GPU needed.
Three engines are available and switchable at runtime: Deepgram (live streaming, ~$0.0043/min, requires a DEEPGRAM_API_KEY), OpenAI Whisper (accurate, ~$0.006/min, requires an OPENAI_API_KEY), and Local Whisper (free, offline, CPU-only, 1 to 4 seconds latency after release). If a cloud engine fails, PipeVoice retries the utterance on Local Whisper automatically.
Yes. The app is MIT-licensed and free to download and run. If you use the Deepgram or OpenAI cloud engines you pay those providers directly with your own API key; costs are roughly pennies per day at typical dictation volumes. The local Whisper path has no recurring cost of any kind.
Yes, that is the primary use case. PipeVoice injects keystrokes into the focused window, terminals included. Per-app profiles let the terminal receive raw text with an automatic Enter press while other windows receive cleaned, non-auto-sent text. The optional MCP server exposes a listen tool that lets MCP-compatible agents like Claude Code prompt you by voice.
No. The cloud engines (Deepgram, OpenAI Whisper) need no local hardware beyond a microphone. Local Whisper runs on CPU out of the box and auto-uses CUDA if present, but CUDA is not required. The local model is ~150 MB, far smaller than GPU-dependent voice tools that require gigabyte-scale downloads.
#1 in the Wispr Flow guide
Cross-platform offline dictation on Windows, macOS, and Linux, with the largest community of the eight
#2 in the Wispr Flow guide
Apple Silicon Mac dictation with local AI enhancement and no subscription
#3 in the Wispr Flow guide
macOS 15+ dictation with live transcription preview and optional on-device AI cleanup
#4 in the Wispr Flow guide
Cross-platform dictation with documented meeting transcription, live diarization, and optional cloud processing
Sponsors
Alternative to
Superhuman
AI email assistant that drafts replies and clears your inbox
Alternative to
Grok Bot
AI teammates you own: your keys, your model, your machine.
Live monitoring on LinkedIn, X, Reddit & more. Agentic social listening for startups.
The projects we add and every new guide. Free, unsubscribe any time.
Claude Code to 
OpenCode
Jira to 
Windshift
Notion to 
AppFlowy
Slack to 
Mattermost
Linear to 
GitLab
Wispr Flow to 
Handy
Arc to 
Zen Browser
CapCut to 
Concat
Grok Bot to 
rakazo
Superhuman to 
Inbox Zero
Airtable to 
AITable
Muse to 
OpenClaw
Railway to 
Coolify
GitHub Copilot to 
Cline
Datadog to 
NetdataConfluence to 
Wiki.js
Microsoft Teams to 
Rocket.Chat
Firebase to 
Supabase
Pinecone to 
Qdrant
Zapier to 
n8n
1Password to 
KeePassXC
Devin AI to 
OpenHands
Asana to 
Leantime
Contentful to 
Strapi
