How can we help?
Troubleshooting guides, common questions, and direct support options for TypeSay.
Start here
Getting Started
Install TypeSay, download a Whisper model, and start dictating in under two minutes. No account or configuration required.
Installation Guide →Report a Bug
Found something broken? Send us a message with steps to reproduce, your OS version, and the Whisper model you were using.
Submit a Report →Frequent Questions
Find answers to the most common questions about TypeSay, from offline use and model selection to refunds and multi-device support.
Browse FAQs →Common issues and fixes
Open TypeSay Settings and navigate to the Audio tab. Confirm that the correct input device is selected from the dropdown. If your microphone does not appear, check that it is connected and recognized by your operating system. On macOS, you may need to grant microphone permission in System Settings > Privacy & Security > Microphone.
Try switching to a smaller Whisper model. The Tiny model (75 MB) transcribes in about 1 second on most hardware. If accuracy is the priority, use the Small or Medium model with GPU acceleration enabled. Ensure your environment has low background noise, and speak clearly at a normal pace. You can also try setting the language manually instead of relying on auto-detection.
Another application may be using the same keyboard shortcut. Open TypeSay Settings > Hotkeys and try assigning a different key combination. TypeSay validates shortcuts in real-time to prevent conflicts. On Linux, make sure TypeSay has permission to register global hotkeys (some Wayland compositors require additional configuration).
TypeSay supports two insert modes: Type Simulation and Clipboard. If one mode is not working with a specific application, try switching to the other in Settings > Output. Clipboard mode copies the text, pastes it via Ctrl+V (or Cmd+V), and then restores your previous clipboard contents. Some terminal emulators and elevated (admin) windows may block type simulation.
Models are downloaded from HuggingFace on first use. Check your internet connection and firewall settings. If the download stalls, close TypeSay, delete the partial model file from the models directory, and try again. The models directory is located in your system's app data folder (e.g., %APPDATA%\TypeSay on Windows, ~/Library/Application Support/TypeSay on macOS).
GPU acceleration requires compatible hardware and drivers. NVIDIA cards need CUDA drivers installed. Apple Silicon Macs use Metal automatically. For Vulkan support, ensure your GPU drivers are up to date. Check the Performance tab in Settings to confirm which backend TypeSay is using. If GPU is available but not active, try restarting the application.
Frequently asked questions
Yes. After the initial model download from HuggingFace, TypeSay operates entirely offline. No audio, text, or metadata is ever sent to any server. The application contains zero telemetry and zero analytics. You can verify this yourself since the source code is open under the MIT license.
For everyday use, the Base model (142 MB) offers the best balance of speed and accuracy. If you need faster results, try Tiny (75 MB). For professional or multi-language work, Small (466 MB) or Medium (1.5 GB) provide higher accuracy. You can switch models anytime from Settings without restarting.
Yes. Your purchase covers all personal devices you own. Install TypeSay on your desktop, laptop, and any other machines you use. There are no device limits and no activation servers.
TypeSay supports 18+ languages through Whisper AI, including English, Spanish, French, German, Italian, Portuguese, Dutch, Russian, Chinese, Japanese, Korean, Arabic, Hindi, and more. You can set a specific language or let TypeSay auto-detect what you are speaking.
Contact us at info@typesay.app within 14 days of purchase for a full refund. No questions asked. Include your order number or the email address used for purchase.
What you need to run TypeSay
| Component | Minimum | Recommended |
|---|---|---|
| OS | Windows 10 / macOS 11 / Ubuntu 20.04 | Latest stable release |
| CPU | 4 cores, x86_64 or ARM64 | 8+ cores for faster transcription |
| RAM | 4 GB (Tiny model) | 8 GB+ (Small/Medium models) |
| Disk | 200 MB (app + Tiny model) | 2 GB (all models downloaded) |
| GPU (optional) | N/A (CPU works fine) | NVIDIA (CUDA), Apple Silicon (Metal), or Vulkan-compatible |
| Microphone | Any input device | Dedicated USB or headset mic |
Contact us directly
We typically respond within 24 hours.
Browse the source code yourself.
TypeSay is MIT licensed. Audit the code, report issues, or contribute on GitHub.
View on GitHub