The interruption when using Deepgram's multilingual transcriber seems to be too sensitive. Anyone experiencing the same problem?