← Back to glossary

Speech interface that captures audio, transcribes it, and processes intents, often via cloud services.

Definition

Voice assistants record audio segments, run speech-to-text, and process intent using models. Some processing occurs on-device; other parts may occur in the cloud with logging and retention.

In plain English Speech interface that captures audio, transcribes it, and processes intents, often via cloud services.

Why this matters

Why it matters: Audio and transcripts can contain personal information; strong consent, minimization, and local processing reduce exposure.

Example

Example: Configure wake-word processing on-device and disable cloud retention of transcripts where possible.