The main technology feature is a real-time singing voice correction, modelling artificial polyphony by using quasi-periodical model with non-stationary parameters based on the instantaneous frequency of the pitch in order to achieve natural sound, low algorithmic delay (<30 ms), a wide range of available voice changing height.
This technology provides an opportunity to develop a real-time singing processor «Sing Assist Karaoke». It processes incoming audio and transforms it to appropriate correct notes. The application synthesizes appropriate voice, preserving original singer's voice, its perception and recognition.
Another options for the technology implementation: