Adaptive Voice Activity Detection

iriSound has a new feature called AVAD

When working in noisy environments the background noise changes constantly. The same does the level of your speech. The human brain simply adjusts the speaking level according to surrounding noise. During the day, and even during a conversation, we constantly adjust our voice to a desired level.
For the very same reason any VHF radio has an adjustable Squelch, that set the noise cut off to a reasonable level. But it only works for the radio – what about each user in a larger intercom setup. Is your intercom VOX operated or a fully open circuit ? Wind noise, engine noise and even breathing, can easily open the microphone, which requires either a constant adjustment of the squelch or the squelch level to be set so high, that very loud speaking is necessary no matter background noise level. AVAD takes care of that. iriSound’s new DSP Software listens and adjusts that for each channel independently. That means in fact that each user in a multiple-user scenario will achieve total silence in between information, and only a slight word will activate and open the microphone circuit.


Monitoring the microphone 6000 times pr. second AVAD Monitors the microphone 6000 times each second to maintain relevant information about the environmental noise as well as the speech level. Constantly, the squelch level will be adjusted to a level where only peaks (voice) will open the microphone. AVAD ensures complete silence in between speech.

Adaptive Voice Activity Detection

AVAD is more than an automatic Squelch Accurate and instant knowledge about the surroundings and the user's speech pattern allow the system to adapt to the background noise and the speech of the user to only transmit when the user is speaking. At low level noise AVAD opens the microphone at a slight whispering and at high volume noise AVAD detects the speech excellently. AVAD has been tested at sea at 40 knots in direct wind.

Adaptive Voice Activity Detection

Adjustable reference level iriSound is pre-delivered with a 30% reference level. That means in fact that speaking must be 30% higher than the background noise, to activate the squelch gate.

In some cases it may be beneficial to have selective levels. In very quit surroundings, where whispering is preferred, a disconnection of AVAD might be best, and in some cases the squelch opens too early according to speech pattern. Then 50% and 70% may be the choice.

By adding an iriSelect this feature is available.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top