My VoIP app has the same issues. Regarding your step 9, there are some other ways for me to increase the volume.: AVAudioSession.sharedInstance().overrideOutputAudioPort(.none) AVAudioSession.sharedInstance().overrideOutputAudioPort(.speaker) or the user changes the output to the receiver and then switches to the speaker (without using .defaultToSpeaker). I think when using .playAndRecord, all sounds will be played with a low volume. Additionally, the Voice Processing IO can further decrease or increase the volume. So, there are three different cases when it comes to volume: Low when using .playAndRecord only Lowest when using Voice Processing IO Highest after step 9 I tried using Remote I/O and my sound always plays at a stable low volume. I can easily fix this issue by changing the sound file. However, I encountered a bigger problem - the volume during conversation is very low. So, I have decided to abandon this solution. To conclude, my main objective is to achieve a stable volume in Voice Processing IO.