OpenWhispr

What is OpenWhispr

OpenWhispr is an open-source dictation and transcription software that runs locally on your device. It uses AI models like OpenAI Whisper and NVIDIA Parakeet to convert speech to text, and also offers meeting note-taking and audio transcription. The core application is free and MIT-licensed, with optional paid cloud services.

How to use OpenWhispr

1. Download and install OpenWhispr for macOS, Windows, Linux, or iOS.
2. Set a hotkey for recording (customizable in Settings).
3. Press the hotkey and speak; your words appear in any text field.
4. For meeting notes, use the AI note taker feature to capture and summarize meetings.
5. Upload audio files for transcription via the audio upload feature.
6. Optionally, configure local models or bring your own API keys for cloud processing.

Features of OpenWhispr

• Local speech-to-text using OpenAI Whisper and NVIDIA Parakeet models.
• 100+ languages with auto-detection and mid-conversation switching.
• System-wide push-to-talk hotkey with customizable key combinations.
• AI agent mode for voice commands like "clean this up" or "draft an email."
• Learns your vocabulary from corrections (names, jargon, terms).
• Works with any application that accepts text input (Slack, VS Code, Chrome, etc.).
• Privacy-first: local processing keeps audio on device; cloud processing deletes audio after transcription.
• Open source (MIT license) with code available on GitHub.
• Cross-platform: macOS, Windows, Linux, iOS.
• Offline capable with local models.

Use Cases of OpenWhispr

• Dictating emails, messages, and documents in any app.
• Taking meeting notes without a bot joining the call.
• Transcribing audio files for notes or records.
• Writing code comments or documentation via voice in IDEs.
• Communicating in multiple languages with auto-detection.
• Using voice commands to format or draft text.

Pricing

• Core app: Free and open source, with unlimited local model usage and bring-your-own API keys.
• OpenWhispr Cloud: Free tier includes 2,000 words/week; Pro plan at $8/month removes the word cap and adds mobile access, meeting recording, and email support.

FAQ

How private is it? Can anyone see my transcriptions?
With local processing, voice data never leaves your device. Cloud processing discards audio immediately after transcription. No data is stored, sold, or used for training. Open source allows verification.

Can I use my own API keys instead of OpenWhispr Cloud?
Yes, you can use API keys from OpenAI, Anthropic, Google Gemini, Groq, or any compatible provider, giving you control over costs and data.

Is OpenWhispr really free?
Yes, the core app is free forever with unlimited local and BYOK usage. OpenWhispr Cloud offers a free tier with 2,000 words/week; Pro ($8/mo) removes the cap and adds features.

What platforms and apps does it work with?
Runs on macOS, Windows, Linux, and iOS. Works with any application that accepts text input, such as Slack, VS Code, Chrome, Word, Notion, and terminals.

What languages are supported?
Supports 100+ languages including English, Spanish, French, German, Chinese, Japanese, Arabic, and more, with auto-detect for switching.

What's the difference between local and cloud processing?
Local processing runs models on your machine (private, offline, hardware-dependent). Cloud processing sends audio to a server (faster, more accurate, requires internet).

How do I customize the hotkey?
In Settings, you can set any key or combination as the recording hotkey, with hold-to-record and push-to-toggle modes.

Is it really open source? Can I contribute?
Yes, it's MIT licensed with source on GitHub. Contributions, bug reports, and feature requests are welcome.