Comment by yNeolh

That happens in most speech to text systems, even Superwhisper, Monologue and Wispr Flow. I read somewhere it comes from training on YouTube audio and happens when there is silence. I guess it depends on the model but most of them are based on Whisper which has this problem