Minimal push-to-talk dictation for Windows and macOS.
Hold a hotkey, speak, release — text appears wherever your cursor is focused.
Polyflo is a lightweight desktop dictation app built with Tauri (Rust backend) and powered by Sarvam AI Saaras v3 speech-to-text. It runs in your system tray and works in any application that accepts typed text.
| Mode | What happens |
|---|---|
| Transcribe | Speech → auto-detect language → paste as spoken |
| Translate | Speech → transcribe → translate to English → paste |
Saaras v3 supports 23 languages (22 Indian languages + English) with automatic language detection and code-mixed speech handling — a strong fit if you dictate in Hindi, Tamil, Bengali, or mix languages mid-sentence.
- Go to Releases on GitHub.
- Download the installer for your platform:
- Windows:
Polyflo_*_x64-setup.exe(NSIS installer) - macOS (Apple Silicon):
Polyflo_*_aarch64.dmg - macOS (Intel):
Polyflo_*_x64.dmg
- Windows:
- Run the installer (Windows) or open the
.dmgand drag Polyflo to Applications (macOS). - Launch Polyflo — it appears in your system tray.
- Right-click the tray icon → Open Settings and enter your Sarvam API key.
First launch: If no API key is configured, the Settings window opens automatically.
Prerequisites
- Node.js 18+
- Rust 1.77+
- Tauri prerequisites for your OS
Windows
- WebView2 (pre-installed on Windows 10/11)
- Microphone access
- Build with PowerShell or CMD — not Git Bash (MSVC
link.execonflict)
macOS
- Xcode Command Line Tools
- Accessibility permission for auto-paste (System Settings → Privacy & Security → Accessibility)
git clone https://github.com/Crisiswastaken/PolyFlo.git
cd PolyFlo
npm installOptional dev fallback: copy .env.example to .env and set SARVAM_API_KEY. Production builds store the key in the OS credential manager via Settings.
# Development
npm run tauri:dev
# Production build (PowerShell/CMD on Windows)
npm run tauri:buildInstallers are written to src-tauri/target/release/bundle/.
Tag a version to trigger the release workflow:
git tag v1.0.0
git push origin v1.0.0GitHub Actions builds Windows and macOS installers and creates a draft release you can review and publish.
- Tray icon — Polyflo lives in the system tray. Right-click for Open Settings or Quit.
- Hotkey — Hold the default shortcut to dictate:
- Windows:
Ctrl+Shift+Space - macOS:
Cmd+Shift+Space - Change it anytime in Settings.
- Windows:
- Speak — A small overlay shows listening/processing state.
- Release — Transcript is pasted into the focused field (or copied to clipboard with a notification if paste is unavailable).
- History — Recent transcripts appear in Settings → Clipboard.
Your Sarvam API key is stored in the OS credential manager (service: polyflo, user: sarvam). It never leaves your machine except when sent to Sarvam's API for transcription. Never commit real keys to git.
Sarvam offers free trial credits with pay-as-you-go pricing at roughly ₹30/hour of audio for speech-to-text — billed per second, rounded up.
Wispr Flow is a polished, subscription-based AI dictation product. Polyflo targets a different trade-off: minimal, open, and pay-per-use via your own Sarvam key. Below is an honest comparison based on each product's public documentation and pricing pages as of 2026.
| Polyflo | Wispr Flow | |
|---|---|---|
| Pricing | Pay-as-you-go via Sarvam API (~₹30/hr STT); free trial credits | Free tier: 2,000 words/week desktop; Pro $12–15/mo unlimited |
| Account | Sarvam API key only | Wispr account required |
| Open source | Yes — build and audit yourself | Closed source |
| Platforms | Windows, macOS (desktop) | Mac, Windows, iOS, Android |
| Languages | 23 languages (22 Indian + English), code-mixing | 100+ languages |
| Indian languages | Core strength — Saaras v3 built for Indic speech | Supported, not Indic-specialized |
| Word limits | None (limited by API usage/cost) | 2,000 words/week on free desktop tier |
| AI editing | Raw transcription + optional English translation | AI auto-edits, filler removal, Command Mode (Pro) |
| Snippets / dictionary | Clipboard history only | Custom dictionary & snippets, synced across devices |
| Privacy | Your key → Sarvam only; lightweight local app | Cloud processing; Privacy Mode on all plans; HIPAA-ready |
| App size | Small Tauri binary (~few MB + WebView) | Larger native + cloud stack |
| Offline | No (requires Sarvam API) | No (cloud-only) |
| Best for | Developers, Indic-language users, cost-conscious heavy dictation, self-hosted control | All-in-one polished UX, mobile, AI formatting, teams/enterprise compliance |
When Polyflo is the better fit
- You dictate primarily in Indian languages or code-mixed speech (Hindi + English, etc.).
- You want no monthly subscription and prefer transparent per-minute API billing.
- You dictate more than ~2,000 words/week and don't want a $12–15/mo Pro plan.
- You want an open, auditable desktop app you can build yourself.
- You need a minimal tray app without account signup beyond an API key.
When Wispr Flow is the better fit
- You want AI cleanup (filler removal, formatting, tone adjustment) and Command Mode voice editing.
- You need iOS/Android or cross-device sync (dictionary, snippets, history).
- You prefer a turnkey subscription with no API key management.
- You need enterprise compliance (SOC 2 Type II, SSO/SAML) out of the box.
Sources: Wispr Flow, Wispr Flow pricing, Wispr Flow docs, Sarvam Saaras v3, Sarvam pricing, Sarvam voice-to-text.
- Tauri v2 + Rust backend
- cpal microphone capture → 16 kHz mono PCM → Sarvam WebSocket STT (
saaras:v3) - Sarvam Translate API for English mode
- enigo paste simulation with clipboard + notification fallback
- Circular always-on-top overlay with live audio waveform
Windows-only helpers for local dev environment setup live in scripts/:
install-windows-sdk.ps1— Windows SDK setupsetup-msvc.ps1— MSVC toolchain setuprun-dev.ps1— convenience dev launcher