Crate oximedia_mir

Source

Expand description

Music Information Retrieval (MIR) system for OxiMedia.

This crate provides comprehensive music analysis capabilities for audio content:

§Tempo and Beat Analysis

Tempo Detection - BPM detection using autocorrelation and comb filtering
Beat Tracking - Beat and downbeat detection with dynamic programming
Onset Detection - Transient detection using spectral flux and HFC

§Tonal Analysis

Key Detection - Musical key detection (Krumhansl-Schmuckler algorithm)
Chord Recognition - Chord progression analysis using chroma features
Melody Extraction - Dominant melody line extraction
Harmonic Analysis - Harmonic-percussive separation

§Structure Analysis

Structural Segmentation - Section boundary detection
Self-Similarity Analysis - Pattern and repetition detection
Section Labeling - Intro, verse, chorus, bridge identification

§High-Level Features

Genre Classification - Genre detection from audio features
Mood Detection - Valence and arousal estimation
Loudness Analysis - Integrated loudness and dynamics

§Low-Level Features

Spectral Features - Centroid, rolloff, flux, contrast
Rhythm Features - Rhythm patterns and complexity
Pitch Features - Pitch class profiles and chromagrams

§Usage

use oximedia_mir::{MirAnalyzer, MirConfig, FeatureSet};

// Create analyzer with default configuration
let config = MirConfig::default();
let analyzer = MirAnalyzer::new(config);

// Analyze audio samples (f32, mono or stereo)
let samples = vec![0.0_f32; 44100]; // 1 second of silence
let sample_rate = 44100.0;

// Perform analysis
let result = analyzer.analyze(&samples, sample_rate)?;

// Access results
if let Some(ref tempo) = result.tempo {
    println!("Tempo: {:.1} BPM (confidence: {:.2})", tempo.bpm, tempo.confidence);
}
if let Some(ref key) = result.key {
    println!("Key: {} (confidence: {:.2})", key.key, key.confidence);
}
if let Some(ref genre) = result.genre {
    println!("Genre: {} (confidence: {:.2})", genre.top_genre().0, genre.top_genre().1);
}

§Patent-Free Implementation

All algorithms are implemented using patent-free methods:

Autocorrelation-based tempo detection
Chroma-based chord recognition
Spectral-based onset detection
Krumhansl-Schmuckler key detection

§Real-Time Capable

Many features support frame-by-frame processing for real-time applications.

Re-exports§

pub use midi::AudioToMidi;
pub use midi::AudioToMidiConfig;
pub use midi::MidiNote;
pub use midi::MidiTempo;
pub use midi::MidiTranscription;
pub use streaming::StreamingAnalysisSummary;
pub use streaming::StreamingAnalyzer;
pub use streaming::StreamingConfig;
pub use streaming::StreamingFrameFeatures;

Modules§

audio_events: Audio event detection (silence, speech, music, applause transitions).
audio_features: Audio feature extraction for Music Information Retrieval.
audio_similarity: Cosine, Earth-mover, and multi-modal audio similarity measures. Audio similarity metrics for Music Information Retrieval.
beat: Beat tracking and downbeat detection.
beat_tracker: Beat tracking and tempo hypothesis scoring.
beat_tracking: Beat tracking using auto-correlation and onset detection.
chord: Chord recognition and progression analysis.
chord_recognition: Chord recognition using chromagram template matching.
chorus_detect: Chorus and refrain detection in musical audio.
chroma_cache: Cached chromagram that amortises FFT cost across multiple consumers. Cached chromagram computation shared between chord recognition and key detection.
chromagram: 12-bin chroma vector computation (pitch-class profiling, chord/key analysis). Chromagram computation and analysis for pitch-class profiling.
cover_art_features: Visual cover-art feature extraction (dominant colours, edge histograms). Cover art visual feature extraction for Music Information Retrieval.
cover_detect: Cover version detection for music information retrieval.
dj_features: DJ workflow features: Camelot wheel, beat-matching, compatibility scoring. Real-time DJ features: beat-matching and compatible key detection.
dynamic_range: Dynamic range analysis — LRA (Loudness Range), crest factor, compression detection.
energy_contour: Short-time energy contour extraction and analysis.
fade_detect: Fade-in and fade-out detection in audio tracks.
fingerprint: Audio fingerprinting module for oximedia-mir.
genre: Genre classification, hierarchy, similarity, and blend detection.
genre_classifier: Gaussian Naïve Bayes genre classifier operating on pre-extracted audio features.
genre_classify: Genre classification for music information retrieval.
genre_classify_new: Neural-network-free rule-based genre classification (spectral features). Fine-grained genre sub-classification from spectral audio features.
harmonic: Harmonic analysis and separation.
harmonic_analysis: Chord recognition and harmonic analysis for Music Information Retrieval.
harmonic_percussive: Harmonic-percussive source separation (median-filter HPSS). Harmonic-percussive source separation (HPSS) for music analysis.
harmonic_spectral: Inharmonicity, HNR, and THD from the harmonic series. Fine-grained harmonic spectral analysis for Music Information Retrieval.
instrument: Fine-grained instrument recognition from MFCC and spectral features.
instrument_classifier: Instrument family and voice classification. Advanced instrument classification from spectral audio features.
instrument_detection: Musical instrument detection from spectral audio features.
key: Key detection using Krumhansl-Schmuckler algorithm.
key_detection: Musical key detection using the Krumhansl-Kessler profile method.
loudness: Loudness analysis and dynamics.
lsh_similarity: Locality-sensitive hashing for fast approximate audio nearest-neighbour. Locality-Sensitive Hashing (LSH) for fast approximate music similarity search.
lyrics_align: Forced alignment of lyrics text to detected onset timestamps. Lyrics timing alignment stub.
melody: Melody extraction and contour analysis.
melody_extract: Energy/zero-crossing melody extraction and contour shape analysis. Melody extraction and contour analysis.
melody_extractor: Harmonic-salience Viterbi melody extractor with vibrato detection. Predominant melody extraction using harmonic summation salience.
midi: Audio-to-MIDI basic transcription.
mir_feature: Generic feature vector utilities for Music Information Retrieval.
mood: Mood and emotion detection.
mood_detection: Music mood and emotion detection using the Russell circumplex model.
multitrack: Multi-stem analysis with per-stem feature extraction. Multi-track / stem-based music information retrieval.
music_summary: High-level music summary — intro/verse/chorus markers derived from structure analysis.
onset_peak: Onset peak picking with configurable threshold and minimum interval. Onset peak detection with adaptive thresholds for music information retrieval.
onset_strength: Onset strength envelope computation for Music Information Retrieval.
pitch_key: Musical key detection using the Krumhansl-Schmuckler algorithm.
pitch_track: Frame-by-frame fundamental-frequency (F0) tracking.
playlist: Music playlist intelligence for oximedia-mir.
playlist_gen: Music playlist generation using musical constraints.
rhythm: Rhythm feature extraction.
rhythm_pattern: Rhythm pattern analysis and drum-pattern recognition.
section_segmenter: Structural section segmentation via novelty-curve self-similarity. Music section segmentation via self-similarity matrix and novelty curves.
segmentation: Music structure segmentation.
similarity: Music similarity and distance metrics.
similarity_search: Brute-force and LSH similarity search over fingerprint indices. Music similarity search using fingerprints and Locality-Sensitive Hashing (LSH).
source_separation: Source separation — vocal / drum / bass / other stem splitting.
spectral: Spectral feature extraction.
spectral_contrast: Spectral contrast feature extraction for music analysis.
spectral_features: Audio feature extraction for Music Information Retrieval.
streaming: Streaming / incremental MIR analysis for real-time use.
structure: Structural segmentation and similarity analysis.
structure_analysis: Music structure analysis: verse, chorus, bridge detection and segment boundaries.
subgenre: Sub-genre tagging within broad genre families. Sub-genre classification extending the top-level genre system.
tempo: Tempo detection and BPM estimation.
tempo_map: Tempo map for music information retrieval.
tempo_stability: Tempo stability analysis (variance, drift, class). Tempo stability analysis module.
thumbnail: Waveform thumbnail generation for audio browsers. Audio thumbnailing — extract the most representative 15–30 second clip.
tuning_detect: Musical tuning detection for audio content.
vocal_detect: Vocal presence detection and vocal/instrumental classification.
watermark: Audio watermark embedding (spread-spectrum LSB steganography). Audio watermark detection using spectral analysis.
watermark_detect: Audio watermark detection and extraction. Simplified spread-spectrum audio watermark detection.

Structs§

AnalysisResult: Complete analysis result.
BeatResult: Beat tracking result.
ChordResult: Chord recognition result.
FeatureSet: Feature set flags for selective feature extraction.
GenreResult: Genre classification result.
HarmonicResult: Harmonic analysis result.
KeyResult: Key detection result.
LoudnessResult: Loudness analysis result.
MelodyResult: Melody extraction result.
MirAnalyzer: Main MIR analyzer.
MirConfig: Configuration for MIR analysis.
MoodResult: Mood detection result.
RhythmResult: Rhythm features result.
SpectralResult: Spectral features result.
StructureResult: Structure analysis result.
TempoResult: Tempo detection result.

Enums§

MirError: MIR error type.

Type Aliases§

MirResult: Result type for MIR operations.

Crate oximedia_mir

Crate oximedia_mir Copy item path

§Tempo and Beat Analysis

§Tonal Analysis

§Structure Analysis

§High-Level Features

§Low-Level Features

§Usage

§Patent-Free Implementation

§Real-Time Capable

Re-exports§

Modules§

Structs§

Enums§

Type Aliases§

Crate oximedia_mir