Syntiant, Sensory deliver multi-language speech recognition

January 08, 2020 //By Peter Clarke
Syntiant, Sensory deliver multi-language speech recognition
AI chip startup Syntiant and software developer Sensory are demonstrating a speech recognition system that supports commands in dozens of languages with the computation done without using the cloud.

The two companies are combining Sensory's wake word engine and voice control with Syntiant’s neural processors to deliver a deep learning, multi-lingual voice interface for battery-powered devices. The combination could also support voice-based user identification.

The NDP100 and NDP101 neural decision processors can support dozens of local voice commands and consume less than 140 microwatts while performing local processing of audio events, Syntiant claims.

Syntiant Corp. was founded in 2017 to develop analog neural networking processors for edge computing. The company has received investment from Microsoft, Amazon Alexa, Applied Ventures, Intel Capital, Motorola and Robert Bosch Venture capital.

Sensory Inc. was founded in 1994 and pioneered neural network approaches for embedded speech recognition in consumer electronics. Sensory's technologies have shipped in over 2 billion products.

Syntiant has also joined the Qualcomm Extension Program, in the hope of improving its ability to design-in touch-free, cloud-free, machine-learning voice interfacesfor always-on speech applications in battery-powered devices, such as Bluetooth-enabled earbuds and hearable devices.

Related links and articles:

News articles:

Startup proposes special SRAM for AI

Microsoft, Alexa, Bosch join Intel by investing in Syntiant

MEMS microphone comes with open DSP

Vous êtes certain ?

Si vous désactivez les cookies, vous ne pouvez plus naviguer sur le site.

Vous allez être rediriger vers Google.