ABR Edge Time Series Processor
Powering the next generation of products
The ABR Edge Time Series Processor (TSP1) is a groundbreaking AI chip that offers unparalleled processing power and efficiency. It is designed to handle complex AI workloads with minimal power consumption, making it ideal for battery-powered devices.
Use Cases
Made with your products in mind
From voice interfaces to smart medical devices and industrial IoT, the TSP1 chip delivers versatile solutions across a wide range of industries.
Voice
Recognition
Biomedical
Monitoring
Industrial
Applications
Specifications
The TSP1 chip features
The Applied Brain Research’s TSP1 is a time series neural network accelerator chip that provides next-level AI capabilities like natural voice interfaces or advanced bio-signal classification in battery-powered devices. The TSP1 provides fully self-contained processing of one or more networks for a wide range of voice or other sensor signal application workloads.
- Power:
- VDD: 1.65-3.6V
- VDD Core: 0.8V (integrated DC-DC supply)
- High-efficiency ABR neural network processing element fabric
- Run powerful signal pattern recognition and other neural networks, enabling lower power, cost, and latency
- Keyword Spotting Trigger Function <2mW
- Full vocabulary speech recognition (speech-to-text) <50mW
- Produces powerful AI inference with low latency (< 20 ms)
- Secure network and firmware on-chip non-volatile storage.
- Supports up to 4 stereo audio inputs and one TDM streaming output
- SPI and I2C master for sensor interfaces
- Multiple programmable GPIO pins
- I2C and SPI target interface for host CPU
- Low-power integrated PMU and clock
- 42-pin WLCSP (0.5mm pitch) or 44-pin QFN package options
