time delay neural network

A feedforward neural networkfeedforward neural network, n.: A neural network laid out like an assembly line: each stage hands its result only to later stages, never back, so every answer depends on the current input alone and nothing is remembered between inputs. for sequential signals in which each unit sees a short window of consecutive time frames of the layer below, with the same weights applied at every time shift (weight sharingweight sharing, n.: Reusing the same learned numbers in many places in a network, for example one edge detector applied at every spot in an image, so the network needs fewer parameters and learns from less data.), while higher layers integrate over progressively longer windows. In modern terms it is a one-dimensional convolutional neural networkconvolutional neural network, n.: A kind of neural network built for images and other grid-like data. It slides small pattern detectors across the input, finding edges first, then shapes, then whole objects., giving shift-invariant recognition of patterns such as phonemes regardless of their exact timing, and it was trained by backpropagationbackpropagation, n.: The method neural networks use to learn from mistakes: it works backward from the error to figure out how much each internal connection contributed, so every connection can be adjusted in the right direction.. TDNN layers later became a standard component of acoustic models in speech recognition and of speaker-embedding networks in the 2010s, and the first Siamese networkSiamese network, n.: A matching machine built from twin copies of one neural network: each copy turns one of two inputs, say two signatures or two face photos, into a summary, and the two summaries are compared., for signature verification (1993), was built from TDNNs.

The acoustic model stacks TDNN layers whose context widens from 3 to 15 frames.