TalkTo is an audio front-end (AFE) algorithm built in DSP Concepts' Audio Weaver platform that delivers exceptionally clean voice signals to voice assistants and ASR engines. It can be integrated with any cloud or edge-based wake word and/or ASR engine, drastically improving VUI performance metrics such as response accuracy rate (RAR) and false rejection rate (FRR). TalkTo can scale from nearfield, low-power, single-mic solutions running on lightweight MCU/DSPs to advanced farfield, multi-mic, multichannel solutions running on powerful SoCs, by using a combination of Audio Weaver technologies, such as AEC, Beamforming, DoA, Noise Reduction, and DSP Concepts' proprietary Adaptive Interference Cancellation, which can eliminate interfering noise without the need of a reference channel.
対応製品 | STM32F4,STM32F7,STM32L4 |
サイト |
Served Countries:
Asia, Europe, North America