This device is not compatible.
You will learn to:
Interact with audio files using a programming language.
Extract different features from the audio files.
Create training and testing datasets from extracted features.
Classify the emotion from audio files.
Skills
Machine Learning
Deep Learning
Prerequisites
Intermediate knowledge of Python
Basic knowledge of scikit-learn
Basic understanding of machine learning models
Technologies
Python
librosa
Scikit-learn
Project Description
Librosa is a Python package that analyzes music and audio. Librosa is mainly used with audio data, such as in music generation (via LSTMs) and automatic speech recognition. It provides the building blocks for developing music information retrieval systems.
In this project, we will explore the Librosa package and perform the following operations:
- Interact with audio files using Librosa.
- Extract the following features of the audio file:
- Mel Frequency Cepstral Coefficients (MFCC).
- Chroma of the audio file.
- Spectral Scale of the pitch of the audio.
After extracting all the features from the audio file, we’ll create different datasets for training and testing. We will then initialize a new classifier using scikit-learn to classify the audio file features to detect emotions. Finally, we will compute the accuracy of our classifier.
Project Tasks
1
Explore Librosa
Task 0: Get Started
Task 1: Import Modules
Task 2: Load Audio and Plot Correlation
Task 3: Plot Multiple Waves
Task 4: Feature Extraction
Task 5: Define Dictionary of Labels
Task 6: Get Labels from Features of Audio Files
2
Create Classifier
Task 7: Import scikit Modules
Task 8: Split the Training and Testing Data
Task 9: Plot the Emotions
Task 10: Initialize the Classifier
Task 11: Classify and Get the Predictions
Task 12: Calculate Accuracy
Task 13: Plot the Loss Curve
Congratulations
Relevant Courses
Use the following content to review prerequisites or explore specific concepts in detail.