Yes This is a common issue, and it’s usually not the voice itself, it’s the gain/normalization in the pipeline.