aalto1 untyped-item.component.html
Low-Latency Pitch-Bending Technique for Audio Signals
Loading...
URL
Journal Title
Journal ISSN
Volume Title
Helsinki University of Technology |
Master's thesis
Electronic archive copy is available via Aalto Thesis Database.
Checking the digitized thesis and permission for publishing
Instructions for the author
Instructions for the author
Location:
Authors
Date
Department
Major/Subject
Mcode
S-89
Degree programme
Language
en
Pages
66
Series
Abstract
A new high-quality low-latency pitch-bending algorithm is presented.
The algorithm consists of three main sections: pitch-shifting algorithm, onset processing algorithm, and gain control and equalization.
The pitch-shifting algorithm is based on the resampling and time-scale modification method.
However, the previous time-scale modification methods could not achieve the required latency and quality for the system.
A new time-domain time-scale modification algorithm has been designed, which is called the Normalized Filtered Correlation Time Scale Modification (NFC-TSM) method.
In order to modify the time-scale of the signal, some parts of the signal are repeated or discarded.
The best splicing point is searched in the normalized low-pass filtered signal using the Average Magnitude Difference Function (AMDF) as the correlation function.
In order to obtain the feeling of realistic playing, the onset processing algorithm is used.
It detects the onset of the input signal with the required latency and performs time-synchronization with the new onset.
Gain control and equalization are used to simulate the gain and timbre of the physically pitch-shifted signal.
The presented algorithm can be exploited in the applications that are using audio effects particularly.
Because of being low-latency, it can be used in real-time performance of music.