logo Ircam

Montage Sound Texture Synthesis

Personnes Associées: Seán O'Leary
Date of Activity: 


Sound Examples for a Montage Approach to Sound Synthesis

Synthesized examples twice the duration of the originals. 

Babble original Montage


original Montage
Bubbling Water original Montage


original Montage
Fire original Montage
Heavy Rain original Montage
Helicopter original Montage
Swamp original Montage
Lapping Waves original Montage
PneumaticDrills original Montage
ShakingPaper original Montage
Sparrows original Montage
Steam Railroad original Montage
Stream Waterfall original Montage
Wind Whistling original Montage

Montage - Removing Unique Elements

This steamrail sample has an audible unique event.  The montage approach is used to identify and replace this unique event.

Steam Railroad original Montage Montage Unique

Montage - Recovering Damaged Textures

This helicopter sample has two missing pieces.  The montage approach is used to identify and replace these pieces.  Notice that the replaced events preserve the quasi-periodic nature of the sample.

Helicopter original Damaged Recovered


Montage and Wavelet Tree Algorithm

The following examples compare results from the wavelet tree algorithm of Dubnov et al. [1] and results from the montage approach to sound texture synthesis.


original Wavelet Montage
Traffic Jam original Wavelet Montage
Baby original Wavelet Montage
Racing original Wavelet Montage

Montage and AudioTexture Algorithm

The next samples compare the montage approach to that of Lu et al.'s AudioTexture method [2].  Notice the periodic tilling in the AudioTexture Rain sample.


original AudioTexture Montage
Stream original AudioTexture Montage


[1] Shlomo Dubnov, Ziv Bar-joseph, Ran El-Yaniv, Dani Lischin- ski, and Michael Werman,“Synthesis of sound textures by learning and resampling of wavelet trees,” 2002


[2] Lie Lu, Liu Wenyin, and Hong-Jiang Zhang, “Audio textures: theory and applications,”Speech and Audio Processing, IEEE Transactions on, vol. 12, no. 2, pp. 156–167, 2004.