How to calculate results about vqmivc

[Wendison]

MOS is 5-scale (1-bad, 2-poor, 3-fair, 4-good, 5-excellent), each listener gives the point for each wav, then we can obtain a series of points given by all listeners for each VC system, the points of each VC system are averaged to obtain the final MOS.

[sauravpd29]

Okay so a listener gives score points out of 5 manually upon listening converted audios. There isn't any other accuracy, precision or something else calculated for comparison?

[Wendison]

Apart from giving the subjective MOS, we also used ASR to test the CER/WER of converted voice and F0-PCC to measure the pitch variation consistency between the converted voice and source voice, more details can be found in our paper.