BTK 2.0: Documentation

Beamforming toolkit provides the C++ and Python libraries for

• Feature extraction,
• Subband processing,
• Binaural processing,
• Speaker tracking,
• Beamforming,
• Post-filtering,
• Dereverberation, and
• Acoustic echo cancellation.

The components of the toolkit are written in C++. Users can build an entire system in C++. In addition to the C++ program interface, the BTK provides Python modules. The BTK is suitable for developing real-time audio and speech signal processing applications.

First Steps

• Getting started
  • Prerequisites for Installation
  • Installing BTK2.0

Docs

• User Guides
  • Feature Extraction
    • Log Power Feature
    • Mel-Frequency Cepstral Coefficient (MFCC)
    • Minimum Variance Distortionless Response (MVDR) Feature
  • Subband Processing
    • Oversampled DFT-modulated Filter
    • Cosine-modulated Perfect Reconstruction Filter
    • Overlap-Add/Save Method
  • Binaural Processing
    • Binaural processing based on Interaural Time Delay (ITD)
    • Energy-based Binary Masking
    • Binary Masking based on interaural intensity difference (IID)
  • Beamforming
    • Data Independent Beamforming
    • SOS Adaptive Beamforming
    • HOS Adaptive Beamforming
    • Spherical Beamforming
  • Array Post-Filtering
    • Zelinski Post-Filtering
    • McCowan Post-Filtering
    • Lefkimmiatis Post-Filtering
  • Dereverberation
    • Subband Weighted Prediction Error (WPE) Algorithm
  • Acoustic Echo Cancellation
    • Kalman Filtering AEC
    • NLMS AEC
  • Time Delay Estimation / Speaker Tracking
    • GCC-PHAT Time Delay Estimation
    • Direct Kalman Filtering Speaker Tracking
    • SRP-PHAT Source Localization
• Developer Guides
• References

About

• About BTK 2.0
  • Contributors
• MIT License