VAD – WhisperTrigger FAQ

WhisperTriggerTM decoded with FAQs

WhisperTrigger™ is an ultra low power, standalone, self-adaptive, Voice Activity Detector (VAD), which can be integrated either within a SoC or in a MEMS microphone, available for both analog and digital interfaces

With the WhisperTrigger the goal is to keep the elements of the audio chain (ADC, Filters, DSP) in an off-mode until a voice is detected. As soon as a voice is detected the WhisperTrigger™ sends an interrupt to wake up the elements in the audio chain. As a result the WhisperTrigger™ offers up to a 90% saving in power consumption. It consumes as low as 10uA.

Yes absolutely, the detection works up to 6-8 meters. With its self-adaptive algorithm, it ensures voice detection in either noisy or quite conditions. Additionally the WhisperTrigger™ algorithm can be programmed by the DSP to ensure the best performance even in the most stringent user conditions.

The WhisperTrigger™ is designed to automatically adjust its detection algorithm to the background noise. This allows the WhisperTrigger™ to have the best rate of true detection, despite the environmental conditions, such a restaurant, and an office or at home. The background noise information or the NFI output (Noise Floor Information) is provided to the DSP to enhance the efficiency of the keyword-spotting algorithm.

The WhisperTrigger ™ offering can be classified in two categories:

  • The WhisperTrigger-a which is dedicated to interface with an analog microphone

The WhisperTrigger can enhance the characteristics of the microphone by adding Voice Activation capability. It allows ultra low power operation by waking up sleeping blocks (e.g. ADC) only when necessary. In specific cases the WhisperTrigger-a along with an operating mode controller (without excess pins), a cap-less voltage regulator and high performances ADC can be embedded within a Digital microphone. 

Truncation or simply cutting a word is a possibility supported by KWS algorithms without interfering with their keyword spotting capabilities. The powerful algorithm of the WhisperTrigger™ ensures the fastest detection even in stringent environments. The evaluations with the MIWOK a public benchmark, shows 100% voice detection with just 30% truncation of Vowel phonemes.

Thankfully WhisperTrigger™ is your best friend which improves power saving up to a 90% and enhances the efficiency of the keyword spotting algorithm by sharing vital Noise floor information.

Dolphin Design has developed a standardized public test bench called MIWOK, which can be freely downloaded from our website and used to check the accuracy of Voice activation in the context of voice first devices. Here is a quick link to MIWOK page which enables to compare three critical parameters: – Voice Detected as Voice – VDV, Noise Detected as Voice – NDV and Detection Latency – DL

Based on a 2850mAh battery of a popular smart speaker, WhisperTrigger™ enhances its battery life from 3 weeks up to 3 years

With a very small silicon footprint and without the requirement of any additional components on the board, integrating the WhisperTrigger™ is like spreading butter on bread.

In case of integration within a Digital Microphone, no additional pin is needed, thanks to the innovative mode controller feature.

Voice assistant such as smart speakers.

All Voice first Devices such as home assistants, Bluetooth wearables, Hearables, smart watch, smart TV, smart Phones, Tablets, Computers etc. Surveillance applications, Car infotainment

We are open to exploring possibilities and welcome any contact in this regard.