omarmhaimdat / SoundClassifier

Building a Sound Classification iOS Application using AI

Home Page:https://heartbeat.fritz.ai/building-a-sound-classification-ios-application-using-ai-1655506d1763

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

A Simple Sound Classifier

Sound Classifier Using Turicreate and CoreML

Open In Colab

We'll create an iOS application that recognize sounds that are listed in the ESC-50 dataset.

Final result

Final Results

Final result - Video with sound

Video with sound - Vimeo

Data preparation

I'm using the ESC-50 dataset.

ESC-50: Dataset for Environmental Sound Classification

I've tested the model on only four classes, even if it's capable of recognizing way more:

  • dog
  • rooster
  • helicopter
  • sneezing

In the python script I've only used the ESC-10 dataset which is way smaller than the ESC-50 one. You can change the script to not exclude the

Train the model

MacOS, Linux & Windows:

Set your python environment and run the following script:

./python > main.py

This model is tested on turicreate v5.7

##Colab Notebook

https://colab.research.google.com/drive/1hKPoEPVP8jGcfC5_mKk9cTWI-Mxvy4zV

About me

Omar MHAIMDAT

Distributed under the MIT license. See LICENSE for more information.

About

Building a Sound Classification iOS Application using AI

https://heartbeat.fritz.ai/building-a-sound-classification-ios-application-using-ai-1655506d1763

License:MIT License


Languages

Language:Jupyter Notebook 79.8%Language:Swift 17.1%Language:Python 3.0%