Abstract | ||
---|---|---|
Recently, Head Related Transfer Function (HRTF) based techniques have been promising approaches for 3D sound synthesis. To ensure high quality of 3D sound synthesis, it requires to utilize personal HRTFs of the listener, which are usually obtained by a complicated and time consuming procedure. To personalize HRTFs, one possibility is to construct multiple linear regression models between anthropometric data and features of HRTFs. Due to high dimensionality of HRTF datasets, it is inefficient to use the original set for the purpose of personalization. To avoid such inefficiency, Principal Component Analysis (PCA) has been proposed to reduce dimensionality of HRTF datasets before customization. Based on the fact that HRTF datasets can be considered as three way data arrays, in this paper we propose three multi-way array analysis methods for HRTF customization. Performance of these three methods is compared with PCA based approaches by several experiments. |
Year | Venue | Keywords |
---|---|---|
2010 | Aalborg | acoustic signal processing,array signal processing,principal component analysis,hrtf based technique,hrtf customization,pca based approach,head related transfer function,multiway array analysis,feature extraction,linear regression,transfer functions |
Field | DocType | ISSN |
Head-related transfer function,Pattern recognition,Computer science,Feature extraction,Curse of dimensionality,Transfer function,Artificial intelligence,Principal component analysis,Linear regression,Personalization | Conference | 2219-5491 |
Citations | PageRank | References |
3 | 0.54 | 5 |
Authors | ||
4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Martin Rothbucher | 1 | 23 | 3.58 |
Marko Durkovic | 2 | 30 | 3.34 |
Hao Shen | 3 | 32 | 6.88 |
Klaus Diepold | 4 | 437 | 56.47 |