bluemacaw.org

What It Is

bluemacaw is an open-source, cross-platform speech-to-text dictation app with a very clear purpose: hold down a hotkey to speak, release it, and the transcribed text is pasted into the app where your cursor is currently active. It supports macOS, Windows, and Linux, and uses a BYOK model: users bring their own third-party speech recognition API keys, while the project itself does not provide any backend service.

Core Capabilities and Models

bluemacaw does not train or host STT models itself. Instead, it comes with integrations for 10 providers: OpenAI, Groq, Grok xAI, Deepgram, AssemblyAI, ElevenLabs, Fal, Gladia, Azure OpenAI, and Rev.ai. The documentation examples mention models such as gpt-4o-mini-transcribe, whisper-large-v3, and nova-3. Its strength is that users can freely switch based on cost, speed, and recognition quality; the limitation is that the overall experience depends heavily on the chosen provider.

Pricing and Trial

The bluemacaw app is free, open-source, requires no account, and has no upgrade tiers or usage limits. The real cost comes from the third-party STT providers. The main text says that various providers offer free or low-cost tiers that are sufficient for testing, but it does not specify exact quotas. The app dashboard can show estimated costs for different providers and models, which is practically useful for long-term dictation users.

Privacy, Integrations, and Ease of Use

Privacy is one of its main highlights: there is no telemetry, analytics, or error reporting; audio is only sent to the provider selected by the user; API keys are stored in the system credential store, such as macOS Keychain, Windows Credential Manager, or Linux Secret Service; transcription text is stored locally in SQLite, with a default rolling retention period of 1 year that can be adjusted. Its integration model is desktop-oriented: record via a global hotkey and automatically paste into the current app. On macOS, microphone and accessibility permissions are required. On Linux Wayland, due to security restrictions, it cannot directly simulate paste actions, so users need to press Ctrl+V manually.

Pros, Cons, and Best Fit

Its advantages are open-source transparency, no proprietary backend, cross-platform support, a wide choice of providers, and controllable costs. The downsides are that users must apply for API keys themselves, which creates a higher onboarding barrier for non-technical users; current Windows builds are unsigned and may trigger SmartScreen; Windows/Linux are marked as Beta, and Linux also has the Wayland paste limitation. It is well suited to developers, writers, privacy-conscious knowledge workers, and anyone who wants to add voice input to arbitrary desktop applications.

Access from China

The main text does not provide information on mainland China network access, payment, or availability, so its access status is unknown. Actual usability also depends on the network connectivity and payment support of the selected STT provider in mainland China. If third-party APIs are unstable, local Whisper clients, MacWhisper, Superwhisper, or official tools from speech recognition providers may be considered as alternatives.