Poboljšanje kvalitete govornog signala primjenom konvolucijskih neuronskih mreža prikladnih za izvođenje na mikrokontroleru

undergraduate thesis

• Title (english): Improving Quality of Speech Signal Using Convolutional Neural Networks Suitable for Microcontroller
• Author: Nika Šljubura
• Mentor: Vedran Bilas (mentor)
• Committee member: Vedran Bilas (predsjednik povjerenstva)
• Committee member: Darko Vasić (član povjerenstva)
• Committee member: Stjepan Begušić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-07-07, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Rad se bavi razvojem neuronskih mreža i implementacijom istih na mikrokontroler iz porodice ARM Cortex-M4 s ciljem poboljšanja kvalitete govornog signala u slušnim aparatima. Radi sigurnosti korisnika, zvukovi hitne buke ne smiju biti eliminirani, zbog čega sustav koristi klasifikator, čija je zadaća prepoznati sadrži li zvučni signal hitnu buku. Ovisno o rezultatu klasifikacije, signal se dalje prosljeđuje jednom od modela s ciljem smanjenja intenziteta odgovarajuće pozadinske buke i povećanja intenziteta govornog signala. Kriterij uspješnosti rada definiran je STOI ocjenom, čiji je iznos jednak 0.86 za signale bez zvukova uzbune i 0.7 za signale koji sadrže zvukove uzbune.

Abstract (english)

The thesis deals with the development of neural networks and their implementation on a microcontroller from the ARM Cortex-M4 family to improve the quality of the speech signal in hearing aids. For the user's safety, emergency noise sounds can't be eliminated, which is why the system uses a classifier, whose task is to recognize whether the sound signal contains emergency noise. Depending on the classification result, the signal is further forwarded to one of the models, which will reduce the intensity of the corresponding background noise and increase the intensity of the speech signal. The success criterion is defined by the STOI rating, which equals 0.86 for signals without alarm sounds and 0.7 for signals containing alarm sounds.

• Keywords: buka; zvuk uzbune; slušni aparat; ugradbeni sustav
• Keywords (english): noise; alarm sound; hearing aid; embedded system
• Language: croatian
• URN:NBN: urn:nbn:hr:168:065831
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-07-13 10:36:07