Data Privacy & Retention Statement
ClearVoice — real-time noise suppression for contact centres
Summary
ClearVoice is architected to process voice audio entirely on-device, within the agent's own browser or desktop application. No audio content is transmitted to, or processed by, any cloud server or third-party service as part of its core noise suppression function.
Is voice/call audio stored or recorded?
No. ClearVoice does not record, capture, or store call audio at any point.
- •Audio is processed in real time using an on-device neural network (WebAssembly-based), running locally in the agent's browser or desktop app.
- •Processed audio is passed directly to the agent's softphone/call platform — it is not written to disk, buffered to a server, or retained after the call ends.
- •There is no cloud round-trip for the noise suppression function: 0 bytes of call audio are sent to the cloud.
Is audio processed by AI, and how?
Yes — but exclusively on-device, not in the cloud. ClearVoice uses a recurrent neural network (RNNoise-based) to distinguish speech from background noise, frame-by-frame, in under 20ms. This AI model runs locally as a WebAssembly module inside the browser/app process — it does not call an external AI/ML API, and no audio is transmitted off the device for this processing to occur.
Is conversation content used for model training?
No. Because no call audio is ever transmitted or stored outside the agent's own device, there is no conversation content available to be collected, reviewed, or used for model training or improvement purposes. This is a structural property of the architecture, not solely a policy commitment.
Data retention — is there a zero-data-retention option?
Call audio retention is zero by default and by design — there is nothing captured to retain in the first place. One distinction worth noting: ClearVoice's optional Usage Analytics feature (used for manager/admin dashboards) stores limited non-audio metadata, such as session start/end times, noise suppression level used, and per-agent adoption/usage statistics. This metadata never includes call audio or conversation content — it is operational/usage data only, used to give team managers visibility into adoption and suppression performance.
Data segmentation & multi-tenancy
ClearVoice uses domain-scoped data isolation: each company's Super User account and agent data is isolated to that company's registered email domain. Companies cannot see or access another company's agents, sessions, or usage data.
Contact
For further questions regarding this statement or a formal Data Processing Agreement (DPA) request, please contact: lance@sonkay.co.za
This statement reflects the technical architecture of ClearVoice and is intended to answer common data-privacy questions raised during vendor evaluation. It is not a substitute for formal legal review. For enterprise procurement requiring a signed Data Processing Agreement or formal compliance certification, please contact us directly.
