IA8201 Enables wake-on-voice processing for low latency voice UI, noise reduction, context awareness, and accelerated machine learning inferencing for edge processing of sensor inputs. The Knowles AISonic Audio Edge Processor IA8201 Open DSP is a high-performance, ultra-low power audio-centric The IA8201, an open DSP platform, brings together leading OpenDSP supporting up to 4 mics, multiple high contributors to the intelligent voice ecosystem to improve audio speed interfaces and GPIOs in two package options performance in a variety of use cases. This Knowles partner (eWLB and QFN). program brings world-class algorithm and cloud contributors to an ecosystem where a multitude of solutions solve complex audio problems, increasing the versatility of an IA8201-based Multi-Core Audio Processing solution. IA8201 combines two Tensilica-based, audio-centric DSP cores one for high power compute and ML, the other for very low Use Case Examples power always-on processing of sensor inputs. Either one can also function as a general-purpose controller using the Tensilica Low Power Voice Wake: Listens for specific OEM keywords to based instruction set. A rich set of audio and general purpose wake the host processor. Large memory enables processing of high speed interfaces enable flexible interfacing with digital multiple stages on-chip for accurate results. microphones, motion, environmental sensors, and as needed Proximity Detection: When combined with an ultrasonic capable with a host for further processing. 1MB of user RAM enables speaker and microphone, detects the distance between the storage of multiple algorithms and voice keywords. system and an object can replace an IR-Prox sensor in bezel-less phones. Highly Optimized, Advanced Instruction Set Hub: Determines location of voice source while tuning out a noisy environment and lowering music to detect voice The DSP SDK (Software Development Kit) with Knowles and commands. Simultaneously takes metadata input and overrides Xtensa HiFi 3 instruction sets enable extensive audio capabilities beamformer to focus on camera-tracked objects. for voice and audio processing, voice user interface, and ambient sound processing. Optimized frame-based processing Security System: Activate with a voice command. Detect glass utilizes floating-point data types, SIMD, and a flexible extended breakage/smoke-alarm, log direction of noise source, trigger instruction set for non-linear functions and accelerated DNN alarm, and send alerts through WiFi connection. MACs. Wireless Earbuds: Delivers low power premium wake-on-voice performance, talk detection to eliminate false triggers, enhanced voice quality through advanced beam forming and noise reduction algorithms as well as support for local commands including answer/ignore calls. Core Details IA8201 Specifications DeltaMax is an Xtensa LX5 128-bit 4-way floating-point DeltaMax, optimized for compute Multi-Core SIMD DSP core, with Knowles instruction set extensions HemiDelta, optimized for low power optimized for high performance frame-based audio Up to 4x PDM Digital Microphones- 1 stereo Audio Interfaces processing and DNN acceleration input, 4x mono inputs, and 1 stereo output, HemiDelta is an Xtensa LX5 64-bit 2-way floating- supporting clock rates up to 6.144 MHz point SMID DSP core, with both Xtensa HiFi 3 and Knowles instruction sets, optimized for low power 2 Up to 3x I S/TDM ports supporting 8 channels each of 32-bit audio data using a 24.576 MHz input clock Additional Features 2 SPI, I C, UART, available GPIOs. Control Interfaces Open DSP IA8201 can be configured as a DSP platform for plugin 1.44MB RAM (1MB available to user) Memory development, without any Knowles Algorithms Contact Knowles Representative for available third 175 MHz Clock party features and algorithms support IA8201BC 1.8V Vdd System Requirements IA8201CQ 1.8V Vdd and 3.3V Vdd Software Capability & Tools -20 to 85C Far-field voice optimization for Mobile and IoT, capable of multiple microphone inputs and stereo AEC Barge-in eWLB 3.00x2.6x0.715mm, 0.4 pitch, 42 ball Packaging Options Capable of on-system keyword trigger processing, hybrid QFN 6.00x6.00x0.75mm, 0.5 pitch, 40 lead second-stage keyword verification, and cloud ASR-A implementation Development libraries for voice communication and For more information, visit www.knowles.com/IA8201 voice interface, including noise suppression, For documents and SDK, sign up at beamforming, echo cancellation, speech enhancement