magistrsko delo
Abstract
Razpoznava govora je sistem, ki omogoča avtomatsko pretvorbo govora v besedilo. Izhod takšnega sistema je surovo besedilo brez velikih začetnic, ločil in ostalih oblikovnih lastnosti. Ker je takšno besedilo nepregledno, ročno urejanje pa zahteva veliko dela, so se uveljavile različne metode, ki omenjene težave rešujejo avtomatsko.
Takšni sistemi lahko temeljijo na različnih metodah, vendar so se v zadnjem času predvsem zaradi dobrih rezultatov uveljavili različni tipi nevronskih mrež. Tako smo v sklopu magistrskega dela implementirali sistem, ki za svoje delovanje uporablja rekurenčne nevronske mreže. Preizkusili smo ga z različnimi vektorskimi vložitvami, kot so GloVe, ELMO in BERT. Implementirali smo tudi spletno storitev, ki omogoča, da sistem enostavno integriramo v različne storitve, kot je npr. že prej omenjena avtomatska razpoznava govora.
Keywords
strojno učenje;nevronske mreže;postavljanje ločil;računalništvo;računalništvo in informatika;magisteriji;
Data
Language: |
Slovenian |
Year of publishing: |
2020 |
Typology: |
2.09 - Master's Thesis |
Organization: |
UL FRI - Faculty of Computer and Information Science |
Publisher: |
[M. Rizvič] |
UDC: |
004.85(043.2) |
COBISS: |
32307203
|
Views: |
1526 |
Downloads: |
226 |
Average score: |
0 (0 votes) |
Metadata: |
|
Other data
Secondary language: |
English |
Secondary title: |
Automatic punctuation in raw word sequences |
Secondary abstract: |
Speech recognition is a system that allows for automatic conversion of speech into written text. Such systems typicaly return raw text without any formatting such as capital letters or punctuation symbols. Because such text is unreadable and it also requires a lot of work to edit manually, various methods have been introduced that solve these problems automatically.
Such systems can be based on a variety of methods. However, due to good results they provide, different types of neural networks are mainly used nowdays. As part of the master's thesis, we have implemented a system that uses recurrent neural network to predict punctuation symbols in raw unpunctuated text. We have tried it with different word embeddings such as GloVe, ELMO and BERT. We have also implemented a web service that allows us to easily integrate the system into various other services, such as automatic speech recognition. |
Secondary keywords: |
machine learning;neural networks;punctuation restoration;computer science;computer and information science;master's degree; |
Type (COBISS): |
Master's thesis/paper |
Study programme: |
1000471 |
Embargo end date (OpenAIRE): |
1970-01-01 |
Thesis comment: |
Univ. v Ljubljani, Fak. za računalništvo in informatiko |
Pages: |
68 str. |
ID: |
11911634 |