Please use this identifier to cite or link to this item: http://dx.doi.org/10.18419/opus-11524
Authors: Serai, Dhiren Devinder
Title: Quantization of automatic speech recognition networks
Issue Date: 2020
metadata.ubs.publikation.typ: Abschlussarbeit (Master)
metadata.ubs.publikation.seiten: 60
URI: http://elib.uni-stuttgart.de/handle/11682/11541
http://nbn-resolving.de/urn:nbn:de:bsz:93-opus-ds-115417
http://dx.doi.org/10.18419/opus-11524
Abstract: Recently, end-to-end neural networks based speech recognition systems have received great interests in the speech community. Although these systems offer a wide range of advantages such as high performance, they often rely on a neural network with a large number of parameters, i.e. large memory footprint and large decoding time. The main goal of this thesis is investigation of quantized neural networks for end-to-end speech recognition systems. Different quantization methods like post training quantization, scalar quantization, iterative product quantization and quantization aware training are investigated in this thesis on ASR model. By doing so, the final system is 3.7 times smaller than the baseline system, faster in decoding with a speedup factor of 1.19 and with almost same level of recognition performance.
Appears in Collections:05 Fakultät Informatik, Elektrotechnik und Informationstechnik

Files in This Item:
File Description SizeFormat 
Serai_Masterarbeit.pdf993,48 kBAdobe PDFView/Open


Items in OPUS are protected by copyright, with all rights reserved, unless otherwise indicated.