| Paper ID | E-3-1.6 |
| Paper Title |
ON THE USE OF THE RELATIVE TRANSFER FUNCTION FOR SOURCE SEPARATION USING TWO-CHANNEL RECORDINGS |
| Authors |
Alice Bates, Daniel Grixti-Cheng, Prasanga Samarasinghe, Thushara Abhayapala, Australian National University, Australia |
| Session |
E-3-1: Speech Separation 1 |
| Time | Thursday, 10 December, 12:30 - 14:00 |
| Presentation Time: | Thursday, 10 December, 13:45 - 14:00 Check your Time Zone |
|
All times are in New Zealand Time (UTC +13) |
| Topic |
Speech, Language, and Audio (SLA): |
| Abstract |
This paper investigates the use of the relative transfer function (ReTF) for source separation. The ReTF is a very useful audio feature as it gives a unique signature of the source’s position, as well as the position of the microphones and the environmental characteristics, such as, room dimensions and reverberation time. ReTFs have been used to localize sound sources but have not been thoroughly investigated for the application of source separation. We present theory for the ReTF between two microphones for a single and multiple sources present. From this theory we propose two source separation algorithms. One which is deterministic and enables the separation of two sources when one or both of their ReTFs is known. The other algorithm uses masking in the time-frequency domain and can be used for separating two or more sources. We also explore the limitations and assumptions of the ReTF and the proposed source separation algorithms. |