Audio artifacts when using OpenVINO music seperation

I have been using the OpenVINO music separation for separating voice and sound effects in an audio book I am remixing.

I consistently get some audio “artifact” when using the effect. It does not occur very often in the audio, but during a long audio book it is fairly annoying. I have attached a short clip of where the effect occurs. Both the raw clip and the clip with the artifact. Interesting the artifact occurs in both the instrumental and vocal track.

Do you have any idea why this happens? Any suggestions how to remove the artifact?

- raw clip
- processed clip