Open sidharthm opened 7 years ago
Sharp observation! There's no plans to incorporate it mostly just out of lack of need, although I agree the code does make it look more like an oversight.
Packet loss in audio is much more sensitive because of the poor-sounding artifacts that it produces, and because generally the WebRTC engine will not reduce audio bitrate like it does with video when data loss is detected. Video packet loss will result in a near immediate drop in bitrate, which is computed as you can see. This is why we just take the minimum of both scores as a sort of "close enough" in the scoring.
I noticed that the Javascript implementation for Video MOS computes packet loss, but doesn't incorporate it in scoring.
Are there plans to use this in the future? If so, do you have any suggestions on how it could factor into the scoring?