This repository contains code for constructing an algorithm that understands simple speech commands, a neural network with EfficientNet architecture is used as an algorithm, trained on ten classes of competition data.
Training practice and an autonomous alternative for voice recognition, for my voice assistant for Raspberry Pi
Available here
Train___EfficientNet.ipynb
This is a pytorch implementation neural net with architecture EfficientNet, by uploading pre-trained weights to the net, you can test it right from google colaboratory. demo.ipynb
Below is a confusion matrix of validation data.
