首页|Reports from Departamento de Informatica y Ciencias de la Computacion Advance Kn owledge in Machine Learning (Real-time impulse response: a methodology based on Machine Learning approaches for a rapid impulse response generation for real-tim e ...)
Reports from Departamento de Informatica y Ciencias de la Computacion Advance Kn owledge in Machine Learning (Real-time impulse response: a methodology based on Machine Learning approaches for a rapid impulse response generation for real-tim e ...)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News-A new study on artificial intelligence is now available. According to news originating from Quito, Ecuador, by NewsRx correspondents, research stated, "Simulation of high-definition binaural room im pulse responses using conventional approaches involves a significant amount of c omputational resources, resulting in high computational time, making these appro aches incapable of performing realtime high quality acoustic virtual reality. T his research implemented a methodology for the rapid impulse response generation using the position of a moving listener inside a fixed sound field." Our news correspondents obtained a quote from the research from Departamento de Informatica y Ciencias de la Computacion: "The rapid generation of the impulse r esponse is performed using its representative compressed dimension, with a small er dimension than the original impulse response, learned by variational autoenco ders and long short-term memory neural networks. First, the methodology selects a representative number of impulse responses covering the area of interest using a reliable room acoustic simulator. Second, it generates a dataset with suffici ent impulse responses uniformly distributed through a data augmentation approach using a modified bilinear interpolation from the impulse responses previously s imulated. Third, it applies an unsupervised model to positionally cluster the im pulse responses to reduce the variability of the impulse responses in the given environment. Fourth, it splits the impulse response into time segments and gener ates a dataset per segment and cluster. Fifth, it trains a variational autocoder with a long short-term memory neural network model for each time segment cluste r of impulse responses to infer the correspondent compressed impulse response pa rt. In summary, the impulse response is generated by assigning the current liste ner position to the corresponding cluster and executing the decoders of the vari ational autoencoders with long short-term memory, trained previously. The findin gs are encouraging; the normalized mean absolute error of the impulse responses gathered by the interpolator and the impulse responses generated by the proposed model is less than 15% in the 88% of impulse respon ses reserved for testing."
Departamento de Informatica y Ciencias d e la ComputacionQuitoEcuadorSouth AmericaCyborgsEmerging TechnologiesMachine Learning
NETL
NSTL
