[Discussion] Quality of part of accompaniment where vocals was removed.

Hello, I'm sorry if there is an alike discussion on that topic, couldn't find one.
I'm trying to remove the vocals from the audio to be left with only background music. And I'm not satisfied with the quality of results I get. The sound is somewhat messy in places where speech was in the accompaniment file.
For example:
source audio: [soundcloud](https://soundcloud.com/nlhlielaa0bt/music-75speech?si=0d62e67dce6d46388f0e59a01b815df5&utm_source=clipboard&utm_medium=text&utm_campaign=social_sharing)
vocals: [soundcloud](https://soundcloud.com/nlhlielaa0bt/vocals-1?si=0d62e67dce6d46388f0e59a01b815df5&utm_source=clipboard&utm_medium=text&utm_campaign=social_sharing)
accompaniment: [soundcloud](https://soundcloud.com/nlhlielaa0bt/accompaniment-1?si=0d62e67dce6d46388f0e59a01b815df5&utm_source=clipboard&utm_medium=text&utm_campaign=social_sharing)
I use just a command form quick start section, nothing special. `spleeter separate -p spleeter:2stems -o output myfile.wav`
I would be very thankful for any recommendations on how to improve the overall quality. Maybe another method or model, or some postprocessing technique.