Hyprwhspr
Native speech-to-text for Linux – fast, private, system-wide dictation on Wayland.
Hyprwhspr is the best native dictation tool for Linux Wayland users, offering top-tier local models and GPU acceleration. Its open-source, private approach makes it ideal for power users, though setup requires CLI comfort. If you value privacy and performance over plug-and-play, it's unmatched on Linux. For beginners on X11 or those needing a GUI, Talon or cloud-based tools may be easier.
- Developers and programmers who dictate code
- Writers and journalists wanting hands-free typing
- Linux power users on Wayland
- Privacy-conscious users requiring local processing
- Users on X11 (Wayland-only)
- Those needing mobile or web support
- Beginners uncomfortable with CLI setup
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Hyprwhspr if you use X11 or a non-systemd display manager, need a graphical setup wizard, or prefer a precompiled binary over building from source.
Optional cloud providers require your own API key and may incur usage charges from Gemini, OpenAI, or ElevenLabs.
Hyprwhspr is free and open-source with no subscriptions, telemetry, or monetization. It's ideal for any user on a budget, especially compared to paid dictation tools like Dragon NaturallySpeaking or cloud-based services that charge per minute.
In short
Hyprwhspr — Native speech-to-text for Linux – fast, private, system-wide dictation on Wayland. Best for Developers and programmers who dictate code, Writers and journalists wanting hands-free typing, Linux power users on Wayland. Free to use.
Viability Score
How likely is Hyprwhspr to still be operational in 12 months? Based on 4 signals — momentum (how recently it shipped), wrapper dependency, revenue model, and web presence.
Last calculated: July 2026
How we score →Key Features
- Local speech-to-text with Parakeet TDT V3, Cohere Transcribe, Whisper family
- GPU acceleration (NVIDIA CUDA, AMD/Intel Vulkan) with auto-detection
- CPU mode via onnx-asr for fast inference without GPU
- Models held hot in memory for instant transcription
- Five recording modes: toggle, push-to-talk, auto, continuous, long-form
- Audio ducking (system volume steps down while recording)
- Paste text anywhere via wtype or ydotool with per-app rules
- Mic OSD themed visualizer overlay for Hyprland/Sway/KDE
- Waybar live status indicator
- Multilingual support with optional translate-to-English mode
- Text processing: word overrides, filler word removal, symbol replacements, custom prompts
- WebSocket streaming via Google Gemini, ElevenLabs, OpenAI
- Works on any Wayland compositor with systemd (Hyprland, GNOME, KDE Plasma, Sway)
- Self-healing: recovers from suspend/resume, mic unplug, keyboard hotplug
- Post-transcription hook to pipe output through shell command
About Hyprwhspr
Hyprwhspr is a system-wide speech-to-text tool for Linux, designed exclusively for Wayland compositors. It enables dictation into any application using local AI models for fast and private transcription. Built for developers, writers, and anyone wanting hands-free typing, it supports five recording modes, GPU acceleration with auto-detection, and hot-memory models for instant transcription. Features include cutting-edge local models like Parakeet TDT V3, Cohere Transcribe, and the Whisper family, with onnx-asr for fast CPU inference. GPU acceleration auto-detects NVIDIA CUDA, AMD/Intel Vulkan, or falls back to CPU. Models stay hot in memory for instant transcription. Recording modes cover toggle, push-to-talk, auto, continuous, and long-form with pause and save. Audio ducking lowers system volume while recording. The tool integrates with Waybar, provides a themed visualizer overlay for Hyprland/Sway/KDE, and injects text via wtype or ydotool with per-app paste rules. Privacy is core: local inference means no data leaves your machine. Optional cloud providers (Gemini, OpenAI, ElevenLabs) are supported with secure credential storage. Hyprwhspr is free, open-source (MIT), with no telemetry or monetization. Compared to other Linux dictation tools, it offers deeper Wayland integration, better model support, and a truly local-first design.
Behind the Verdict
Hyprwhspr fills a long-standing gap on Linux: a fast, private, Wayland-native dictation tool that doesn't sacrifice accuracy. Its strength lies in local-first design—models like Parakeet TDT V3 and Cohere Transcribe run hot in memory, delivering near-instant transcription even on CPU via onnx-asr. GPU acceleration auto-detects your hardware, and you can unload models from VRAM when needed. The five recording modes cover everything from brief commands to long lectures, with audio ducking and per-app paste rules adding polish. Privacy is genuine: no telemetry, no cloud dependency by default, and credentials for optional cloud providers are stored securely. The tool is free, MIT-licensed, and well-maintained. Weaknesses: it's Wayland-only (no X11 support), CLI-driven (no graphical interface), and requires comfort with the terminal and package managers like yay or git. Beginners may find the setup barrier steep. It also lacks mobile or web support. For developers, writers, and privacy-conscious Linux users on Wayland, Hyprwhspr is unmatched.
Researching Hyprwhspr? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Hyprwhspr actually fits — and what changes day-one when you adopt it.
You're writing Python functions in VS Code. With a hotkey, you toggle recording, speak your code, and Hyprwhspr injects the text directly into the editor using wtype. Models stay hot for instant transcription.
Outcome: Faster code entry without repetitive strain, hands-free workflow.
You use continuous mode with auto-flush on silence. As you dictate paragraphs, Hyprwhspr transcribes them in near real-time and pastes into LibreOffice. Audio ducking reduces music volume while you speak.
Outcome: Efficient dictation with minimal interruptions, improved throughput.
You configure local inference with Whisper on CPU. You press push-to-talk to capture action items. No audio leaves your machine. Post-transcription hook pipes output to a shell script that saves to a file.
Outcome: Secure, private note-taking with full control over data.
Use Cases
- Dictate code and commands hands-free in your IDE or terminal
- Transcribe meeting notes or journal entries into any text field
- Control your system with voice via custom scripts and hotkeys
- Translate spoken input to English in real-time for multilingual work
- Automate text input in applications without typing, using continuous mode
Models Under the Hood
as of 2026-07-17
Limitations
- Hyprwhspr requires a Wayland compositor with systemd; X11 is not supported.
- GPU acceleration is auto-detected but depends on compatible drivers.
- The tool is primarily CLI-driven; no graphical settings interface is provided.
as of 2026-07-06
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Hyprwhspr tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Free & Open Source
$0
Ideal for
Anyone who wants free, private, local dictation on Wayland Linux — developers, writers, privacy advocates.
What this tier adds
Starting tier: completely free with no subscriptions, telemetry, or monetization; includes all features.
Where the pricing makes sense
The company stage and team size where Hyprwhspr's pricing actually pencils out — and where peers do it cheaper.
Hyprwhspr is free and open-source with no subscriptions, telemetry, or monetization. It's ideal for any user on a budget, especially compared to paid dictation tools like Dragon NaturallySpeaking or cloud-based services that charge per minute.
Setup time & first value
How long it actually takes to get something useful out of Hyprwhspr — broken out by persona, not the marketing-page minute.
For Arch Linux users with yay: install the package and run setup — about 5 minutes. For other distros: clone the repo, run install-deps.sh, then setup — about 15 minutes. Configure models and hotkeys in the config file — another 10 minutes. Total: 10-25 minutes depending on experience.
Switching to or from Hyprwhspr
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From Talon Voice: switch to Hyprwhspr for deeper Wayland integration and local models; retrain custom commands.
- →From Dragon NaturallySpeaking via Wine: migrate to native Linux dictation; no Wine overhead, better model support.
- ↗To Talon Voice: if you need X11 support or a more mature voice control ecosystem.
- ↗To cloud-based dictation (e.g., Otter.ai): if you require web/mobile access and can accept privacy tradeoffs.
Integrations
Resources & Guides
Official links
Tools that pair well with Hyprwhspr
Common stack mates teams adopt alongside Hyprwhspr, with the specific reason each pairing earns its keep.
Featured Head-to-Head Comparisons
Alternatives to Hyprwhspr
View all
Typeless for iOS
AI voice dictation that polishes speech into text 4x faster than typing
Speech to Text Dictation AI
Real-time speech-to-text dictation and translation on iPhone and iPad.
Speechmatics
Low-latency speech-to-text for multilingual, multi-speaker conversations.
Frequently Asked Questions
Categories
Best-of guides
Used Hyprwhspr? Help shape our editorial sentiment research.