Abstract:
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an encoder neural network, a decoder neural network, and a prior neural network, and using the trained networks for generative modeling, data compression, and data decompression. In one aspect, a method comprises: providing a given observation as input to the encoder neural network to generate parameters of an encoding probability distribution; determining an updated code for the given observation; selecting a code that is assigned to an additional observation; providing the code assigned to the additional observation as input to the prior neural network to generate parameters of a prior probability distribution; sampling latent variables from the encoding probability distribution; providing the latent variables as input to the decoder neural network to generate parameters of an observation probability distribution; and determining gradients of a loss function.
Abstract:
A system, method, and apparatus for detecting drones are disclosed. An example method includes receiving a digital sound sample and partitioning the digital sound sample into segments. The method also includes applying a frequency and power spectral density transformation to each of the segments to produce respective sample vectors. For each of the sample vectors, the example method determines a combination of drone sound signatures and background sound signatures that most closely match the sample vector. The method further includes determining, for the sample vectors, if the drone sound signatures in relation to the background sound signatures that are included within the respective combinations are indicative of a drone. Conditioned on determining that the drone sound signatures are indicative of a drone, an alert message indicative of the drone is transmitted.
Abstract:
An embodiment of the invention relates to a method for processing an input signal (S) and generating an output signal (Sout) based on the input signal (S), said method comprising the steps of: In an analysis stage, extracting a plurality of kernels from the input signal (S), wherein each kernel is described by a parameter vector that is defined by a given number of extracted kernel parameters, and forming the output signal (Sout) based on the extracted kernel parameters.
Abstract:
Dynamic loudness equalization of received audio content in a playback system, using metadata that includes instantaneous loudness values for the audio content. A playback level is derived from a user volume setting of the playback system, and is compared with a mixing level that is assigned to the audio content. Parameters are computed, that define an equalization filter that is filtering the audio content before driving a speaker with the filtered audio content, based on the instantaneous loudness values and the comparing of the playback level with the assigned mixing level. Other embodiments are also described and claimed.
Abstract:
A microphone unit has a transducer, for generating an electrical audio signal from a received acoustic signal; a speech coder, for obtaining compressed speech data from the audio signal; and a digital output, for supplying digital signals representing said compressed speech data. The speech coder may be a lossy speech coder, and may contain a bank of filters with centre frequencies that are non-uniformly spaced, for example mel frequencies.