BelaSigna R281

www.onsemi.com

7

System Performance

Recognition Rate

Performance has been measured in various noise

conditions and is shown in Figure 5. Tests were performed

in a sound isolation booth with a series of pre−recorded

trigger and non−trigger phrases playing continuously, mixed

with varying levels of noise. Recognition rate is 100% in a

quiet environment (no noise playing). The system was

trained in a quiet environment. In all cases, the false trigger

rate (the number of times the system triggered on a

non−trigger phrase) was 0%.

Figure 5. Recognition Rate Versus SNR

Current Consumption

Average current consumption was also measured with the device operating in a pure noise environment with no speech

present. The results are presented in Table 9.

Table 9. AVERAGE CURRENT CONSUMPTION

Condition

Average Current Consumption (uA), VBAT = 1.8 V

Quiet Environment

160

Babble Noise

170

*Not including microphone current consumption

Single User Operation

Because BelaSigna R281 performs recognition based on

a training procedure performed by a specific individual,

recognition

of

the

trigger

phrase

is

effectively

“single−user”. Only the person who trained the system (or

an individual with an extremely similar voice print) will be

able to reliably trigger the device.

Digital Interfaces

Wake Up Pin

The Wake Up pin is a digital output (referenced to VDDO)

which is used to indicate that the trigger phrase was detected.

It is intended to be connected to a digital input of an external

host, and its behavior (active high versus active low and

For more information refer to AND9267/D.

Boot Select Pin

This pin is reserved for future purpose and must be

connected to VDDA.

Inter

can be used for high−speed transmission of data between

BelaSigna R281 and an external device. The interface

operates at speeds up to 100 Kbit/sec, and always operates

in slave mode at an address of 0x62.