Privacy Policy
Last updated: 2026/5/16
Effective date: 2026/5/16
Short version: Geaso is a voice input tool. Geaso records microphone audio only after you intentionally start a recording. To turn speech into text, the beta/public version sends recording audio to Geaso's cloud speech backend and may process it through speech recognition or text cleanup providers. We do not sell your data, use it for ad tracking, or use your audio or transcripts to train models.
1. Overview
Thank you for using Geaso. This Privacy Policy explains how Geaso handles microphone audio, transcripts, local history, the cloud speech backend, diagnostic data, and third-party services.
Geaso is designed around one clear path: start recording, transcribe speech, and insert the text into the current app. We try to process only the data needed for that path.
2. Data Geaso Handles
Geaso may handle the following data:
- Microphone audio: voice audio captured after you intentionally start recording.
- Transcript text: partial text, final text, raw text, and optional text cleanup output.
- Target app information: the foreground app name, Bundle ID, or process identifier before recording, used for automatic insertion and diagnostics.
- Diagnostic metadata: task ID, trace ID, recording duration, server duration, error category, service status, charged seconds, and remaining quota.
- Local user dictionary: terms, common misrecognitions, and correction rules you add to improve recognition output.
- License and device data: invite token, token prefix, device ID, and app version for beta access, quota checks, and abuse prevention.
3. When Recording Happens
Geaso captures microphone audio only after you start recording through the global hotkey, floating panel, or menu bar. Geaso does not keep the microphone open for background listening, and it does not upload ambient audio before recording starts.
You can revoke microphone permission in macOS System Settings. If permission is revoked, Geaso cannot record.
4. How Data Flows
In Remote mode, Geaso captures audio on your Mac and sends it to Geaso's cloud speech backend. The backend may forward audio to speech recognition providers, such as Volcengine/Doubao, iFlytek, or similar speech recognition services used in the future.
If text cleanup is enabled, or if server-side post-processing is configured, the final transcript may be sent to an OpenAI-compatible LLM service to lightly clean punctuation, spacing, casing, and obvious recognition errors. This processing should not change the meaning of the transcript.
Apple mode uses macOS Speech as an on-device recognition path. Data handling in that mode depends on Apple's system capabilities and Apple's applicable policies.
5. Local Storage
Geaso stores local History on your Mac so you can review recent recordings, troubleshoot problems, and recover text. Local History may include audio files, transcript text, raw text, task IDs, trace IDs, target app details, audio diagnostics, failure categories, and event timelines.
You can delete individual records or clear all records in Geaso's History, and you can manage local retention in the app. Local deletion affects records on your device. To request deletion of server-side diagnostic records, contact us using the email below.
6. Server Retention
To support troubleshooting, quota settlement, service reliability, and abuse prevention, Geaso's cloud backend may temporarily retain task metadata and necessary transcript text. The default maximum retention period is 7 days.
By default, the server does not retain recording audio long term. Audio or more complete diagnostic material is retained only when you explicitly submit feedback, diagnostic material, or a support request that includes it.
7. What We Do Not Do
- We do not sell your audio, transcripts, or personal data.
- We do not use your audio or transcripts for ad tracking.
- We do not use your audio or transcripts to train models.
- We do not capture microphone audio before you intentionally start recording.
- We do not store third-party ASR credentials in the Geaso client.
8. Third-Party Services
Geaso may use third-party services for speech recognition, text cleanup, cloud infrastructure, email support, and crash or diagnostic support. Speech recognition provider examples include Volcengine/Doubao and iFlytek. Text cleanup may use an OpenAI-compatible LLM service.
We try to send only the data needed to provide the requested service. If provider categories or processing practices materially change, we will update this policy.
9. Security Measures
We take reasonable measures to protect data, including encrypted transport, data minimization, access controls, server-side token hashing, client-side Keychain storage for sensitive tokens, local deletion controls, and short diagnostic retention. No internet service can promise absolute security, but we avoid collecting unrelated data and limit access where possible.
10. Your Choices and Deletion
- You can revoke microphone, speech recognition, or accessibility permissions.
- You can switch to Apple mode when supported to reduce reliance on Geaso's cloud speech backend.
- You can disable text post-processing to avoid sending final text to an optional LLM cleanup service.
- You can delete recordings and transcripts from local History.
- You can contact greatwhole90@gmail.com to request deletion of server records associated with your feedback or diagnostic request.
11. Children's Privacy
Geaso is intended for general macOS productivity use and is not directed to children. If you believe a child has provided personal information without guardian consent, please contact us.
12. Changes to This Policy
If Geaso's data handling, backend topology, third-party services, or retention periods change, we will update this page and revise the date above.
13. Contact
If you have questions about this Privacy Policy or data deletion requests, contact greatwhole90@gmail.com.