What is the Audio Improvement metric?

How do we calculate the IRIS noise improvement score?

Our noise improvement score is then calculated using the difference between SNR (Signal to Noise ratio) of the original call audio^* compared to a theoretical silent call (no background noise to be removed). Signal-to-noise ratio (SNR) is an industry standard measurement of how much signal there is compared to background noise. The bigger this value, the better your SNR, indicating a better quality of the audio (less noise).

From our IRIS SDK (embedded in the IRIS Clarity app), we track the dBFS (decibel full scale, the loudness of digital audio) values of all audio before and after it is processed by Clarity. This is what you see in our audio event charts. 

From these values, we can calculate an average value of the Signal (the cleaned up audio), and an average value of the Noise (the noisy audio removed from the audio stream). Comparing the difference in SNR of the audio compared to silence (-90dBFS), allows you to see how much the quality of the call audio has been improved.

* No audio is sent to IRIS. The Clarity app calculates the loudness of the audio only. We cannot recreate actual audio from this data.