Crate stratum_dsp 

Stratum DSP

A professional-grade audio analysis engine for DJ applications, providing accurate BPM detection, key detection, and beat tracking.

Features

• BPM Detection: Multi-method onset detection with autocorrelation and comb filterbank
• Key Detection: Chroma-based analysis with Krumhansl-Kessler template matching
• Beat Tracking: HMM-based beat grid generation with tempo drift correction
• ML Refinement: Optional ONNX model for edge case correction (Phase 2)

Quick Start

use stratum_dsp::{analyze_audio, AnalysisConfig};

// Load audio samples (mono, f32, normalized)
let samples: Vec<f32> = vec![]; // Your audio data
let sample_rate = 44100;

// Analyze
let result = analyze_audio(&samples, sample_rate, AnalysisConfig::default())?;

println!("BPM: {:.2} (confidence: {:.2})", result.bpm, result.bpm_confidence);
println!("Key: {:?} (confidence: {:.2})", result.key, result.key_confidence);

Architecture

The analysis pipeline follows this flow:

Audio Input -> Preprocessing -> Feature Extraction -> Analysis -> ML Refinement -> Output

See the module documentation for details.

Re-exports

pub use analysis::result::AnalysisResult;

pub use analysis::result::AnalysisMetadata;

pub use analysis::result::BeatGrid;

pub use analysis::result::Key;

pub use analysis::result::KeyType;

pub use analysis::confidence::AnalysisConfidence;

pub use analysis::confidence::compute_confidence;

pub use config::AnalysisConfig;

pub use error::AnalysisError;

Modules

analysis

Analysis and result aggregation modules

config

Configuration parameters for audio analysis

error

Error types for the audio analysis engine

features

Feature extraction modules

preprocessing

Audio preprocessing modules

Functions

analyze_audio

Main analysis function