Voice dictation appsChecked November 2, 2026

TRWhisper

TRWhisper is a free, open-source dictation app for Windows 11 under MIT. It runs Whisper on the device through Whisper.net, with optional CUDA 13 acceleration on NVIDIA GPUs. It is tuned for Turkish with vowel-harmony-aware normalization, and connects to Ollama, Gemini, OpenAI, Groq, DeepSeek, Claude or any OpenAI-compatible endpoint for cleanup. One caveat: Windows 11 x64 only.

OSIAlternative to Wispr Flow, Superwhisper

What does TRWhisper do?

TRWhisper is a push-to-talk dictation app built for Windows 11. The README calls it "an alternative to Superwhisper tuned for Turkish dictation." Press and hold Right Ctrl by default, speak, and release to transcribe into the active app. Toggle mode and a hands-free mode with silence detection are also supported, and the hotkey is fully customisable.

Transcription uses an in-process Whisper.net C# engine, with Whisper CLI alongside. Models are downloaded, switched and deleted from the Settings UI with real-time speed tracking and SHA-256 validation. Silero VAD skips non-speech audio to avoid phantom text. For Turkish, rule-based normalization handles spoken numbers, percentages, currency, dates, times and measurement units, and respects Turkish vowel and consonant harmony when attaching suffixes to numbers.

Key facts

  • Licence: MIT.
  • Platform: Windows 11 x64. .NET 9 is required to build from source. The installer is a single file about 51 MB.
  • Speech engine: in-process Whisper.net with optional Whisper CLI. CUDA 13 acceleration on NVIDIA GPUs needs driver 580.00 or newer; large-v3 Turbo processes dictations in about 2 to 3 seconds on GPU against about 23 seconds on CPU, by the README.
  • LLM providers for cleanup: Ollama (local), Google Gemini, OpenAI, Groq, DeepSeek, Anthropic Claude (default claude-haiku-4-5), and any OpenAI-compatible endpoint such as LM Studio or OpenRouter.
  • Modes: Clean (removes fillers like ııı, eee, şey), Business Email, Bulleted Summary, Code & Technical, English Translation and an AI Assistant mode that answers the dictated instruction instead of cleaning the text.
  • Active window detection activates a matching persona for VS Code, Visual Studio, Outlook, Thunderbird and Windows Terminal.
  • Dictionary: user-defined jargon replacements with Turkish-suffix preservation (pitonda to Python'da).
  • API keys are encrypted with Windows DPAPI using application-specific entropy; keys are never written in plain text or sent over unencrypted HTTP.
  • Daily API call cap for cloud providers; default 200 requests, with Block or Warn-only behaviour at the cap and a warning at 80 percent.
  • Test suite: 420 automated unit tests and a UI smoke tool (uismoke).

What does it replace, and where does it fall short?

TRWhisper replaces Superwhisper and Wispr Flow for a Windows 11 user who wants a local dictation app, especially a Turkish speaker, under MIT. It appears on the open-source alternatives to Superwhisper and Wispr Flow pages.

Where it falls short: Windows 11 x64 only. Spoken-punctuation mode converts standalone nouns too, so "bu nokta önemli" is converted; the feature can be turned off. CUDA 13 is required for GPU speed, which pins TRWhisper to recent NVIDIA drivers. The number and date normalization is Turkish-only; other languages fall back to raw Whisper output. Wispr Flow and Superwhisper have hosted products with support that TRWhisper does not offer.

How does TRWhisper run?

Download TRWhisper-Setup-x.y.z.exe from the releases page (about 51 MB) and run it. No administrator password is required. For CUDA acceleration, install an NVIDIA driver of 580.00 or newer. To build from source, install the .NET 9 SDK.

Settings has tabs for General, Audio, Model & AI, Hotkey, Dictionary, Overlay Pill and Backup & History. A memory card in the Model tab shows whether the model is loaded, which language is active, RAM usage and whether GPU (CUDA) or CPU is in use, with a button to unload the model.

Who is TRWhisper for?

A Turkish speaker on Windows 11 who wants local push-to-talk dictation with correct Turkish numbers, dates and suffixes, under MIT. For a Mac dictation app with the same push-and-paste shape, see Handy; for a Thai-English dictation app, see OLIV. For a hosted product, Superwhisper or Wispr Flow is the paid tool.

What limits does the README state?

  • Windows 11 x64 only.
  • CUDA 13 acceleration needs NVIDIA driver 580.00 or newer.
  • Spoken punctuation also converts standalone nouns such as "bu nokta önemli"; the feature can be turned off.
  • Number and date normalization only runs for Turkish output.
  • Cloud providers for cleanup are optional; a daily call cap is enforced to control cost.
  • An API key reminder fires once a configurable rotation threshold (default 90 days) is crossed.

Questions people ask

Is TRWhisper open source?

Yes. The LICENSE at the repository root is MIT, an OSI-approved licence. The README says TRWhisper is a completely local, zero-telemetry application. It uses Whisper.net, which is open source, and models are downloaded from Hugging Face and validated with SHA-256.

Does TRWhisper work offline?

Transcription is local. The README says "powered by local Whisper. Your voice data never leaves your computer." LLM cleanup is optional and can either use a local provider (Ollama, LM Studio) or a cloud provider (Gemini, OpenAI, Groq, DeepSeek, Claude). With a local provider, nothing leaves the machine.

What does TRWhisper need to run well?

Windows 11 x64 and .NET 9 to build. For speed, an NVIDIA GPU with CUDA 13 support (driver 580.00 or newer); large-v3 Turbo takes about 2 to 3 seconds on GPU and about 23 seconds on CPU by the README. The installer is about 51 MB; Whisper models are downloaded through the Settings UI with speed tracking and SHA-256 validation.

How does TRWhisper compare with Superwhisper?

Superwhisper is a paid macOS product. TRWhisper is a free, MIT Windows 11 app that runs Whisper locally through Whisper.net, with Turkish-specific normalization and a provider-neutral LLM cleanup step. It covers a different operating system and a different language focus. The open-source alternatives to Superwhisper page sets it beside Handy, VoiceInk and others.

How does TRWhisper store my API keys?

API keys are encrypted with Windows DPAPI (Data Protection API) using application-specific entropy (TRWhisper_DPAPI_Entropy_v2). The README says keys are never stored in plain text or transmitted over unencrypted HTTP. A rotation reminder fires once a configurable threshold (default 90 days) is crossed, with one-click deletion and a direct link to the provider's key console.

Sources

More voice dictation apps

ProjectReplacesOpennessStarsLast releaseSelf-host
Handy

A free, offline push-to-talk dictation app for macOS, Windows and Linux that runs Whisper or NVIDIA Parakeet on your own machine and pastes the text at your cursor, under MIT.

OSI
37,047
Nov 2, 2026
Yes
FluidVoice

A GPL-3.0 macOS dictation app that runs Nemotron, Parakeet, Cohere Transcribe, Apple Speech or Whisper on the device, with command and write modes, per-app prompts, and an optional closed local enhancement model.

OSI
12,668
Oct 22, 2026
Yes
OpenWhispr

A dictation and meeting-notes app under MIT that runs Whisper or Parakeet locally or cloud models with your own key, detects Zoom, Teams and FaceTime calls, and labels speakers on device.

OSI
10,547
Oct 29, 2026
Yes
VoiceInk

A macOS dictation app under GPL-3.0 whose README names Superwhisper and Wispr Flow as the tools it replaces, with local models, per-app modes, a personal dictionary and an AI assistant mode.

OSI
7,003
Sep 28, 2026
Yes
Whispering

A speech-to-text app inside the Epicenter monorepo under AGPL-3.0 that records, transcribes with a provider you choose, cloud or local, optionally polishes the text, and pastes it at the cursor.

OSI
4,839
Dec 27, 2025
Yes
Jot

An Apache-2.0 macOS dictation app that calls Gemini 3.5 Transcribe with your own key, writes audio to disk from the first millisecond for crash safety, and inserts polished text at your cursor without an account.

OSI
554
Aug 27, 2026
Yes

Added November 2, 2026. Every claim above comes from the project's README, LICENSE or model card, read on November 2, 2026, or from the GitHub API on the date shown in the panel. Found an error? Write to reveneau@licheo.com and it is fixed in the next weekly pass. Repository: github.com/kagangungor/TRWhisper.

Want TRWhisper running on your own servers?

Reveneau sets it up on your hardware, connects it to your product, and builds the software around it.

Start a project