On-Device Processing
Definition
On-device processing is the practice of executing AI workloads — inference, preprocessing, and postprocessing — entirely on the user's device. For speech recognition, this means the audio captured by the microphone is processed by a locally-running model, and the resulting transcript never leaves the device.
Ummless uses on-device processing for speech-to-text via Apple's built-in speech recognition framework, which runs on the device's Neural Engine. This approach guarantees that raw audio data is never transmitted over the network, providing strong privacy protection. On-device processing also enables offline functionality and eliminates latency from network round-trips.
Frequently Asked Questions
Why does on-device processing matter for voice apps?
On-device processing ensures that your spoken audio never leaves your computer, protecting sensitive conversations, proprietary information, and personal communications from exposure.