Clicking audio artifacts in Streaming API demo

Hello,

when I am using the Streaming API with any of the Multilingual voices, I receive an annoying clicking/tick sound in the audio mainly when a consonant occurs. The amount of clicks and the frequency they occur in vary from voice to voice, but are always present to some extend.
I am using the provided Streaming API demo from GitHub. The text I want the avatar to repeat (not OpenAI) is mostly in German, with a few English words.

From the sample audio of the voices I noticed that some actors did forcefully exhale in the microphone at the end of some words, could this issue come from the training of the voices? I notice the audio error with both headphones and speakers. Could it be related to the WebRTC handling of audio?

Curiously, I don't experience these audio artifacts when using German voices instead of Multilingual.

Thanks in advance.