Multiple fundamental frequency estimation based on generative models

Signal Processing Methods for Music Transcription

Volumn , Issue , 2006, Pages 203-227

  Davy, Manuel ^a  

^a ECOLE CENTRALE DE LILLE   (France)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 75149194552     PISSN: None     EISSN: None     Source Type: Book    
DOI: 10.1007/0-387-32845-9_7     Document Type: Chapter

References (6)

1. 1, is different from the fundamental frequency F0, which is the inverse of the acoustic waveform period.
2. This can be obtained by choosing the AR coefficients whose corresponding characteristic polynomial has zeros with relatively small amplitude [339].
3. It is much more difficult to evaluate the reconstruction accuracy with a psychoacoustically relevant measure. The residual total energy may yield such a performance estimator, though.
4. The probabilities in (7.44) are indicative, and may be adjusted at will, in particular when J(n) reaches its minimum/maximum value.
5. The reader interested in particle filtering may refer to Chapter 2 for a short introduction, and to [152] for a full survey.
6. LN(a; μ, Σ) is a log-Gaussian distribution' means that log(a) is Gaussian with mean log(μ) and covariance matrix Σ.