Constant Q Transform – A Visual Guide
4 days ago
- #audio analysis
- #music processing
- #Constant-Q Transform
- The Constant-Q Transform (CQT) uses logarithmic frequency bins matching human pitch perception, unlike the FFT's linear bins.
- Low-frequency bins have long windows for good frequency resolution, while high-frequency bins have short windows for good time resolution.
- CQT bins are spaced logarithmically, aligning with musical notes (e.g., 12, 24, or more per octave), making pitch analysis more natural.
- Each bin's window length is computed as Nk = Q · fs / fk, ensuring a constant number of cycles per window regardless of frequency.
- Efficient CQT algorithms rely on the FFT for computation, reducing the naive O(KN) complexity per frame.
- Folding CQT bins into a single octave produces a chromagram, useful for chord recognition and key detection.
- CQT is widely applied in music information retrieval, audio synthesis, deep learning, and real-time audio processing.