Fluent.ai 为Cadence Tensilica HiFi 5 DSP 真无线立体声产品提供嵌入式语音识别方案

January 26 2023 -- Fluent.ai announced an embedded voice recognition solution for True Wireless Stereo (TWS) and other hearable products running on the Cadence Tensilica HiFi 5 DSP in partnership with Airoha, a MediaTek subsidiary. The solution is part of an integration available on DSP Concepts’ Audio Weaver platform, which provides OEMs a rapid way to design and prototype audio features and capabilities.

Fluent.ai is a Canadian speech recognition software company founded in 2015. Fluent.ai stands-out in the industry for delivering small footprint embedded software solutions providing offline, noise robust, speech recognition that can support any language or accent. The latest software solution allows users to trigger more than 30 TWS actions through 101 natural and flexible voice commands, including activating noise cancellation, controlling music, accepting or declining calls and checking battery level. Fluent’s patented acoustic-only approach is upending the way people experience voice-enabled listening with hands-free, embedded offline voice recognition that eliminates the need for a companion smartphone app or cloud-based voice assistant.

"The complexity and requirements of TWS products continue to evolve and Fluent.ai is proud to further its development by offering our low-latency, embedded voice recognition solution on Cadence Tensilica HiFi 5 DSP-based Airoha SoCs, a leading platform for TWS products," says Probal Lala, CEO of Fluent.ai. "Our technology, along with the DSP Concepts Audio Weaver platform, will allow OEMs the ability to create fully customizable, intricate and remarkable audio experiences for TWS and other hearable products, now with robust voice command capabilities."

Audio Weaver is the audio development platform that powers sound and voice features in many of the world’s prestigious consumer electronics and automotive brands. Audio Weaver provides product makers with hundreds of building blocks and product-ready algorithms in a graphical, modular design environment that runs on a processor-agnostic embedded framework.

"Audio Weaver allows audio product makers to engineer their products quickly and efficiently, and Fluent.ai’s embedded voice recognition enables OEMs to stay ahead of the curve as customer demand for lower latency, greater privacy, and natural voice UI interaction increases," says DSP Concepts CEO Chin Beckmann.

The Tensilica HiFi 5 DSP smoothly blends neural networks with traditional digital signal processing to efficiently execute audio pre-processing, keyword detection and more natural speech recognition. The Tensilica HiFi 5 DSP includes dedicated MACs and instructions to enable rich voice-controlled user interfaces and AI-based audio applications.

"The Tensilica HiFi 5 DSP is ideal for product makers looking to include differentiating features that may be of high computational complexity in their TWS offerings," says Yipeng Liu, product marketing group director of audio/voice DSPs at Cadence. "The Tensilica HiFi 5 DSP’s low-power design enables OEMs to employ Fluent.ai’s embedded voice recognition solution in a variety of products without sacrificing high computational throughput for real-time audio processing while enabling long battery life."

w.dspconcepts.com
www.cadence.com
www.fluent.ai