This work presents an experience in using neural networks in the area of speaker recognition, in the two variants of verification and identification. The developed system has a classical architecture in the area. It uses blocks for feature extraction, comparision, references and decision. (...)Keywords: speaker identification; speaker verification; filter bank; neural networks; gamma model; gamma memory; f...
Main motivation for this work was the attainment of synthesis with natural quality. One of the promising paths to reach this objective is the use of articulatory synthesis, technique based on direct modellation of the human speech production process. Articulatory synthesis can be used in simulations, and as a versatile informant to produce stimuli for perceptual tests, indispensable in speech pro...
| Financiadores do RCAAP | |||||||
|
|
|
|
|
|
||