Política de privacidade do Tacet

This privacy policy explains what happens to your data when you use Tacet, the desktop speech-to-text app published by Revolt Bots LLC (“we”, “us”). It covers the app only. Our website has its own privacy policy.

The short version

• Tacet has no accounts, no analytics, no telemetry, no crash reporting, no ads, no in-app payments and no automatic update checks.
• We receive no data from Tacet. Nothing in the app sends anything to a Revolt Bots server.
• With the local engine (the default), your speech is transcribed on your own computer. No audio or transcript text is sent anywhere.
• If you choose OpenAI or Deepgram instead, your recorded speech goes directly from your computer to that provider, under your own API key and that provider’s terms.
• Everything Tacet stores (settings, transcript history, usage counts) stays on your computer, in the folder where Tacet is installed.

Transcription engines

You choose the engine in Tacet’s settings.

Local engine (default)

Audio is transcribed on your computer with the open-source faster-whisper models. Your audio and transcripts do not leave your machine.

The local engine does make one kind of network request: it gets its speech model from Hugging Face. The model is downloaded the first time you use a model size, and each time a model is loaded (when Tacet starts or when you change the model) the underlying library checks Hugging Face for that model. This request reveals your IP address and the versions of the libraries in use to Hugging Face. If you have saved a Hugging Face access token on your computer for other tools, the library may send it too. No audio or text is included. If you are offline, Tacet uses the model it has already downloaded.

OpenAI and Deepgram (optional)

If you set the engine to OpenAI or Deepgram, Tacet sends your recorded speech to that provider to be transcribed:

• Audio goes straight from your computer to the provider. We are not an intermediary and never see it.
• The request carries your own API key, the selected model and, if set, your language.
• With live preview turned on (the default), Tacet sends short overlapping clips of what you are saying while you speak, and then the complete utterance, so the same speech can be sent more than once.
• For OpenAI, the server address can be changed in Tacet’s configuration file. If you change it, your audio and key go to the address you configured instead.

What the provider does with your audio is governed by its own terms and privacy policy, not by us. See OpenAI’s privacy policy and Deepgram’s privacy policy.

API keys

Tacet reads API keys only from environment variables (OPENAI_API_KEY and DEEPGRAM_API_KEY by default). It does not save keys to disk and sends a key only to the provider it belongs to.

What Tacet stores on your computer

All of this lives in the folder where Tacet is installed, except temporary audio and the model cache.

What	Contents	How to control it
Settings (config.json)	Your preferences, voice commands, word replacements and templates. It holds the name of the API-key environment variable, not the key.	Settings window, or edit the file
Transcript history (sessions/)	The full text you dictated in each session, with timestamps and word counts. On by default. Tacet keeps the latest 50 sessions.	Delete single sessions or clear all in the History window
Usage statistics (stats.json)	Totals of words, characters, sessions and time. No transcript text.	”Reset stats” in the Statistics window
Temporary audio	Short audio files created while a clip is being transcribed	Deleted automatically after each transcription
Clipboard	Your final text is copied to the system clipboard after each utterance. On by default.	Turn off in Settings
Speech models	Downloaded model files, in the Hugging Face cache folder on your computer	Delete the folder to remove them

Tacet also types the transcribed text into whichever window is active, which is its purpose.

Microphone and keyboard

• Microphone: Tacet opens the microphone when it starts and keeps it open while it runs, so your operating system may show the microphone as in use. Audio is discarded unless you are actively dictating.
• Keyboard: Tacet listens for its hotkey (and, if you enable it, for any key press to stop dictation). It does not record, log or store your keystrokes.

Data we collect

None. Tacet does not send us any data, and we have no way to see what you dictate or how you use the app. If you contact us through our support form, the website privacy policy covers what you send there.

Open source

Tacet is open source under the MIT License. You can review exactly what it does at github.com/revoltbots-com/tacet-voice.

Children

Tacet is a general-purpose tool and is not directed at children under 13. Because we collect no data through the app, we do not knowingly collect personal information from anyone, including children.

Changes to this policy

If a future version of Tacet changes how it handles data, we will update this policy and the date at the top of this page.

Contact

Revolt Bots LLC, 3330 Fairchild Gardens Ave #32704, Palm Beach Gardens, FL 33410, USA. Questions about this policy: use our support form and choose “Tacet - Support”.