music_feature_analyzer library

Music Feature Analyzer Package

A comprehensive music feature analysis package that extracts detailed musical features and metadata from audio files using YAMNet AI model and signal processing.

Platform Support: This package supports Android and iOS only. Desktop (Windows, Linux, macOS) and Web platforms are not supported.

Features:

AI-powered instrument detection
Genre classification
Mood analysis
Tempo detection (BPM)
Energy analysis
Spectral features
Vocal detection
Complete metadata extraction (title, artist, album, album art, etc.)

Usage:

import 'package:music_feature_analyzer/music_feature_analyzer.dart';

// Initialize
await MusicFeatureAnalyzer.initialize();

// Extract metadata
final song = await MusicFeatureAnalyzer.metadata('/path/to/song.mp3');
print('Title: ${song.title}');
print('Artist: ${song.artist}');
print('Album: ${song.album}');

// Extract features
final features = await MusicFeatureAnalyzer.analyzeSong(song);
print('Genre: ${features.estimatedGenre}');
print('Tempo: ${features.tempoBpm} BPM');
print('Instruments: ${features.instruments}');

Classes

Complex

Complex number class for FFT calculations (SAME AS ORIGINAL)

ExtractedSongFeatures

Comprehensive song features extracted by the analyzer

FeatureExtractor

Main feature extractor service

IsolateFeatureData

Data model for passing feature extraction data to isolates This allows us to pre-load assets in the main thread and pass them to isolates

IsolateInputData

Input for background isolate: file path, duration, and pre-loaded model/labels. Pre-extracted audio is passed from main isolate because FFmpegKit (platform channel) cannot run inside compute isolate.

MusicFeatureAnalyzer

============================================================================ MUSIC FEATURE ANALYZER - Main Package Entry Point

Peak

Peak data class for tempo detection (SAME AS ORIGINAL)

SignalFeatures

Signal processing features

SignalProcessingResults

Signal processing results class (same as original)

SongModel

Song model for the music feature analyzer package Contains all metadata information extracted from audio files

YAMNetLabel

YAMNet label with comprehensive categorization (SAME AS ORIGINAL)