Institutional Repository of Vilnius University
Kalbos aptažinimas su Raspberry Pi mini kompiuteriu /
Abstract
dc:descriptionThe aim of this thesis "Speech Recognition with Raspberry Pi Single-Board computer" is to create speech recognition system which can recognize limited number of words using minimal size data set. This system is meant to run on embedded devices in order to follow latest technology trends in edge computing and privacy. The pipeline of this system consists of three major steps. In the first step, voice recording is converted to mono format and down sampled. The second step is a transformation from voice signal to Mel-frequency cepstral coefficients. These coefficients are then used in the third step, where they are fed to convolutional neural network. The output of the network is a vector which values are probabilities of word classes. Classification of a word is made by choosing the maximum probability. The main result of the thesis is over 90% classification accuracy on test data set of the system described above. The system can be deployed on an embedded device and still classify a word in less than 0.25 seconds. This low latency enables it to be used in production environments for speech recognition, for example in automotive cabin or in smart home devices.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2019
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Venclovas, Paulius,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD193353215&prefLang=en_US