|
|
|
|
| Image-Based Audio Processing |
|
|
Sunics technology transforms sound to image - and vice versa. Sounds can be viewed, recognised, enhanced, modified and even synthesised via their corresponding image.
The examples on this page illustrate how image processing turns into audio processing. The images show the wave shapes of Sunics. Playback of the corresponding audio signal is started via mouse hover.
|
|
|
| Noise Removal: irregular background noises are easily removed by removing visual noise.
A simple way to accomplish this is to apply a'blur' effect to the wave shape image.
This removal of visual noise removes the background noise in the corresponding audio signal
|
|
|
|
|
Example 1: an opera singer disturbed by a train noise in the background before (left) and after (right) applying a 'blur' effect to the waveshape.
|
|
|
|
| Sound Source Separation: sounds mixed together can be separated again.
Selecting one sound component and applying above removal process extracts the selected component.
Selection is accomplished via appropriate choice of pitch.
|
|
|
|
|
Example 2: tuning into the vocals of a music piece (left) and applying a 'blur' effect to the waveshape extracts the vocals from the piece (right).
|
|
|
| Audio Synthesis: create an image - create a sound.
Each image corresponds to a unique sound.
Transforming an image that was created via synthesis or capture to its corresponding sound therefore produces an entirely new sound.
|
|
|
|
|
|
Example 3: a drawing (left), fractal (center) and synthesised pattern (right) are each transformed to their corresponding audio, generating a variety of sounds.
|
|
|
|
| Modification: cut/copy/paste can be used to replicate/remove sound features.
Audio features appear as distinct image features. Cutting/copying/pasting these features allows to move, replicate or remove the corresponding audio features.
This example also illustrates how in contrast to traditional tools, modifications can be performed in time and phase direction: selecting a region in the wave shape image corresponds to selecting several successive intervals in the time domain. Cutting/copying/pasting this region then applies to all these intervals simultaneously.
|
|
|
 |
|
|
 |
|
|
 |
|
Example 2: an audio feature of the original sound (top) is replicated (center) and removed (bottom) by replicating/removing the corresponding image feature. The narrow red region in the wave shapes (left column) corresponds to several intervals in the time-waveform (right column).
|
|
|
| More Information
|
|
|
|
|
|
|
|
|
|
