Voice Activity Detection
Ultimate Power Savings
with WhisperTrigger
WhisperTrigger allows your system to detect voice activity and increase its battery life. Our robust and silicon-proven, Voice Activity Detection, provides ultra-low-power Always-On Voice listening IP for a fast time to market.
Ultimate Power Saving
The WhisperTrigger is a stand-alone feature. it enables to switching off the audio DSP and ADC, to divide the power consumption by up to 50 times compared to conventional approaches at a system level
Self-adaptability
WhisperTrigger‘s self-adjustable algorithm ensures the best detection rate whatever the environmental conditions of the user
High Performances
WhisperTrigger‘s outstanding performance – up to 100% detection rate both near field and far field conditions. Available in Analog and Digital
Breakthrough Voice Activity Detection for Ultimate Power Saving
The breakthrough WhisperTrigger tracks continuously the surrounding sounds to identify a voice activity so as to awaken the voice processing blocks (audio ADC and DSP) through a trigger signal as soon as but not sooner than, a voice is detected. Two types of WhisperTrigger (WT-d and WT-a), are proposed for integration into the SoC to support both analog/electric and digital microphones. WhisperTrigger can also be embedded into a DMIC.
WhisperTrigger – analog (WT-a): The ultimate power saving solution
The WT-a enables to switch off the whole chain of acquisition and processing of the audio signal from the ADC to the DSP and to wake them up only when voice is detected by the voice activity detection IP.
WhisperTrigger – digital (WT-d): The fully synthesizable VAD
The WT-d detects the presence of a voice at the output of a DMIC or at the output of a voice ADC embedded on SoC so as to switch on the DSP, to perform Keyword Spotting, only once a voice is detected.
Detection Latency (DL)
WhisperTrigger - d: Down to 4.6 ms
WhisperTrigger - a: Down to 3 ms
Noise Detected as Voice (NDV)
WhisperTrigger- d : Down to 7%
WhisperTrigger - a: Down to 3.6%
Power Consumption of the WT
WhisperTrigger - d : 25 µA
WhisperTrigger - a: 10 µA
MIWOK, an Objective Benchmark for Voice Activity Detection
Dolphin Semiconductor has developed MIWOK, the first public domain benchmark for voice activity detection. It characterizes the performance in different conditions and environments. You can benchmark:
- the detection latency (DL)
- the radio of voice detected as voice (VDV)
- the radio of noise detected as voice (NDV)
