Use an unoptimised local channel, between the main logic and the party in question.
However, in my view, playing background sounds is a bad idea. If nothing else, check your compliance with accessibility legislation, as people with hearing loss can find it particularly difficult to separate the background from the speech. This is a common complaint by older viewers of TV programmes, with background music.
I, from a capable person, had an opensip server made for me, which records a voip trunk and when a call arrives it mixes the VoIp RTP flow with a wav file and then redirects it to an Asterisk PBX..
So I was able to put a file with the classic noise of an office at work and the information processing times (latency) are less annoying.
Thanks for answer @simeone686, its not flexible for me because of I’m receiving audio from provider as chunks and its resource consuming process to join dfifferent part of office voice to little chunks.