Georgios Karakasidis's Projects
Repository for the ''Accent Adaptation Through the Use of Synthesized Speech'' paper, where accent-specific ASR training is assisted with an augmented accented dataset.
Dockerfile for kaldi-gstreamer-server.
Trying to solve the drug (entity) identification task with LLMs.
A python module for performing image segmentation using the EM algorithm, implemented in Rust through PyO3.
End-to-End Speech Processing Toolkit
Config files for my GitHub profile.
Code for creating a GMM-UBM model in python (GMM training is done with Rust). After training the universal GMM, MAP adaptation is performed in order to get the specific GMMs.
Grapheme to Phoneme and Digit to Word conversion for Greek.
A simple gnome shell extension for home assistant.
A Home Assistan systray application.
A minimal, single column latex template for CVs/resumes.
A set of my micropython scripts.
Multilingual TTS data augmentation through a pivot language for low-resource ASR
Modules to convert numbers to words. 42 --> forty-two
Convert numbers to their corresponding Greek word.
Repository for the phonebook exercises of the Fullstack Web-Dev course
Rust Word Error Rate Calculator
Building simple speech recognition systems with kaldi and pytorch-kaldi.
A simple neural network with 1 hidden layer in python
A list of freely available datasets in Greek. I also plan to add some helper scripts for creating your own ASR models.
A PyTorch-based Speech Toolkit
A gnome extension for turning my TV on and off. This is done by using an API I have built on a NodeMcu board which is connected to a relay that controls the power to my TV.
Project Von Muziris as part of the course "AI/ML/DL for industry" at Aalto university