Voice acting of characters not accepting tone/volume prompts (Seedance 2.5)

When giving lines to characters and adding in tone/volume direction (in a whisper, hushed and intimate, etc.), the model seems to ignore these and has everyone speaking in a matter of fact tone and normal conversational volume. Seedance 2.0 is able to use these kinds of directions quite well. It’s odd that an updated model takes a step back on this.