July 10, 2012, Seoul – Spansion (NYSE: CODE), a leading provider of flash memory solutions for the embedded market, announced the Acoustic Coprocessor (ACP), the industry's first Human-Machine Interface (HMI) coprocessor capable of voice control system interfaces. This new product utilizes voice recognition technology from Nuance Communications, the company that provided the voice recognition technology for Apple's Siri. Spansion's Acoustic Coprocessor is ideal for voice recognition systems in applications such as automobiles, gaming, and home appliances, and significantly improves response times and accuracy compared to existing voice interfaces by supporting a vast acoustic database. Furthermore, composed of custom-designed logic and high-speed memory, it accelerates and optimizes the speed of HMIs requiring voice recognition technology while eliminating the audio processing workload typically handled by CPUs.

As electronic products become increasingly AI-driven and support interactive functions, various human interfaces—such as voice, image, and gesture—must more rapidly meet consumer demands for faster recognition speeds and quicker access to relevant information. To date, user interfaces have relied solely on generic hardware and software. Spansion’s Acoustic Coprocessor utilizes Nuance’s VoCon® software engine and adopts the first application-specific approach integrating custom-designed logic and hardware to provide dedicated resources for natural user interfaces, such as voice.

“Spansion has integrated Nuance’s voice recognition technology with a proprietary method that can provide a highly influential platform to automotive manufacturers,” said Michael Thompson, Vice President and Head of Nuance’s Mobile Business Unit. “Specific coprocessors can make voice interactions very fast, more natural, and more accurate.”
Glenda Dorchak, Senior Vice President and Head of Global Business at Spansion, stated, “As human-machine interaction rapidly evolves, the need for more natural user interfaces is emerging, with voice technology leading this trend. Spansion’s new acoustic coprocessor products strengthen our strategy by realizing value-added system performance. The realization of value-added system performance enables customers to drive product differentiation and motivates the widespread adoption of connected AI devices. Through our leadership in automotive and embedded memory, we are designing new system platforms capable of rapidly reading and analyzing data to realize a new dimension of user experience.”

Chris Schreiner, Director of Automotive Consumer Insights at U.S. market research firm Strategy Analytics, stated, “Technological advancements in voice recognition that bring us closer to natural language understanding are important in the automotive market. An increasing number of automakers are integrating voice recognition interfaces into vehicles to support not only mobile connectivity management but also driver safety. Performance improvements, such as accuracy and latency, achieved through technological innovation like Spansion’s Acoustic Coprocessor will provide significant benefits to consumers and bring about a major change in how we interact with our vehicles.”

Spansion's Acoustic Coprocessor leverages Spansion's proven technology and leadership to enable instantaneous response speeds, high reliability, and performance in a wide range of embedded applications. Benchmark results using Spansion's Acoustic Coprocessor show that system response speed is improved by up to 50% compared to using the application processor alone. It reduces the load on the application processor by up to 50%, allowing it to freely perform other tasks. In addition, Spansion's acoustic coprocessor can support a vast database including multiple languages, male/female voice differences, and tonal sounds requiring pitch variations. In other words, this improves accuracy and natural language understanding performance.

Michael Palma, Senior Research Analyst for Semiconductors at IDC, stated, “In today’s computing devices, CPUs handle everything from high-speed internet access and HD video to 3D navigation and voice recognition, leading to bottlenecks in both processing performance and memory bandwidth. Using application-specific coprocessors to manage acoustic grades can significantly improve system responsiveness and latency in voice recognition systems, freeing up more capacity for the CPU to focus on other resource-intensive applications. This also enables a more enhanced user experience across the board.”
The initial implementation of this product took place in the automotive market. In the automotive market, voice recognition technology is being presented as a safer and smarter way to connect with vehicles, minimizing the distraction caused by visual manuals on portable devices. The Spansion demo platform integrates Spansion's technology with Nuance's voice recognition software and is currently being evaluated by major automotive manufacturers. Design samples are expected to be available in the third quarter of this year.