24-01-2015, 04:09 AM
l want to full seminar report and ppt on voice morphing
|
voice morphing seminars ppt free download
|
l want to full seminar report and ppt on voice morphing
29-03-2017, 03:11 PM
Voice transformation (also known as voice transformation and voice conversion) is the alteration generated by the software of a person's natural voice. The purpose may be to add audio effects to the voice, hide the identity of the person or impersonate another person. There are many applications that can benefit from this type of technology. For example, a TTS system with integrated speech morphing technology can produce many different voices. In cases where the identity of the speaker plays a key role, such as dubbing of films and television programs, the availability of high-quality voice morphing technology will be very valuable, allowing the generation of the appropriate voice (perhaps in different languages) Without the original actors being present.
There are basically three interdependent problems that must be resolved before constructing a voice morphing system. In the first place, it is important to develop a mathematical model to represent the voice signal so that synthetic speech can be regenerated and prosody can be manipulated without artifacts. Secondly, the various acoustic signals that allow humans to identify the loudspeakers must be identified and extracted. Third, the type of conversion function and the method of formation and application of the conversion function must be decided.
The aim of this research is to develop flexible algorithms of high quality that can transform the speech of a speaker. A system based on a synchronous sinusoidal step model has been developed that uses LSF character encoding and linear transformations. To ensure high quality, a number of novel techniques have been developed to minimize artifacts that typically result from the loss of glottal source information, formant bandwidth widening, phase incoherence, and spectral colour of non-voiced sounds.
There are basically three interdependent problems that must be resolved before constructing a voice morphing system. In the first place, it is important to develop a mathematical model to represent the voice signal so that synthetic speech can be regenerated and prosody can be manipulated without artifacts. Secondly, the various acoustic signals that allow humans to identify the loudspeakers must be identified and extracted. Third, the type of conversion function and the method of formation and application of the conversion function must be decided.
The aim of this research is to develop flexible algorithms of high quality that can transform the speech of a speaker. A system based on a synchronous sinusoidal step model has been developed that uses LSF character encoding and linear transformations. To ensure high quality, a number of novel techniques have been developed to minimize artifacts that typically result from the loss of glottal source information, formant bandwidth widening, phase incoherence, and spectral colour of non-voiced sounds.
|
« Next Oldest | Next Newest »
|
