This mostly builds on the keyframing we already set up for video, but the audio renderer will now appropriately updated keyframe inputs per sample in accordance with keyframe values.