What's New

• UAR feature is included. UAR stands for usage auto reporting. An API key is required for operation. The SDK usage in minutes can be monitored using https://sdk.krisp.ai dashboard

Bundled Models

• krisp-ac-multi-3.9.1.kef multi accent, multi voice model, robust to audio codec and telephony conditions. The model fits for the inbound/downlink audio use case.
• krisp-ac-phi-server-3.7.0 single voice Filipino accent model.
• krisp-ac-sa-server-3.9.0 single voice South African accent model.
• krisp-ac-latam-server-3.7.0 single voice Latin American accent model
• krisp-ac-uk-server-3.9.1.kef single voice "UK" accent model, optimized for Indian, Filipino and African accents.

Single Voice Accent Models

• All single voice accent models include Krisp BVC (background voice suppression) and background noise removal.
• All krisp-ac-accent-version single accent server models are optimized for earbuds, headset, handset and other near-field audio capturing applications.
• Does not support far-field audio capture

Multi Voice Multi Accent Model

krisp-ac-multi-3.9.1.kef

• Includes background noise removal

• Codec robust and optimized for inbound telephony

• Does not remove audible background voices by design

• Can be used for both near-field and far-field

• Can be used to process multi voice audio received from a conference calls or similar use cases