Deep Speaker Embedding with Multi-Part Information Aggregation in Frequency-Time Domain for ASV