Analysis/Synthesis and Spatialization of Environmental Sounds
This
webpage illustrates a paper submitted to Icad'09. Three sounds
illustrate the analysis/synthesis and spatialization processes. Each
sound is resynthesized three times with different transformations. For
the air swishing sound, the original version is cut into pieces
that are positioned dynamically and played randomly. For the wind and
ocean wave sounds, the resynthesized versions are given different
spatial widths.
Analysis parameters: Na is the analysis window size, La
the analysis hope size. We use different hope sizes for the three
sounds, leading to different compression ratio. In each case, the
stochastic modeling allows saving a significant amount of data compared
to the original signal. Compression ratio is 8 for the air swishing
sound, 32 for the wind and 64 for the ocean wave.
Synthesis
parameters: for all sounds, the synthesis window size is 1024 and the
synthesis hope size is 256. Gains are interpolated when the synthesis
window size is different from the analysis window size (i.e., for the
wind and ocean wave sounds).
All soundfiles can be downloaded in a zip file here Sounds are binaural so please use headphones
Air swishing sound compression: 8 (Na = 1024, La = 256) 00 original 01 synthesis 1 02 synthesis 2 03 synthesis 3
Wind sound compression: 32 (Na = 4096, La = 1024) 00 original 01 synthesis 1 02 synthesis 2 03 synthesis 3
Ocean wave sound compression: 64 (Na = 8192, La = 2048) 00 original 01 synthesis 1 02 synthesis 2 03 synthesis 3