HMM_backend< Obs > Class Template Reference

Backend for a Hidden Markov Model object, templated over the observation type. More...

#include <Markov.hpp>

Inheritance diagram for HMM_backend< Obs >:

Classes
class	State

Public Member Functions
std::vector< std::pair < ChordProgression, Score > >	getBestProgressions (Melody M, unsigned n=1) const
	Return a list of chord progressions with scores. More...
unsigned	getMaxProgressionsNbr () const
void	saveToFile (const std::string &fname) const
	Save the class to a given file. More...
void	learn (Database &db, const std::string &style)
void	learn_unsupervised (Database &db, const std::string &style)

Protected Member Functions
void	forwardProcedure (const std::vector< Obs > &o)
void	backwardProcedure (const std::vector< Obs > &o)
void	updateProcedure (const std::vector< Obs > &o)
void	generateObsSet (const std::vector< Data > &entries)

Additional Inherited Members
Public Types inherited from HMM
using	Proba = double
	Return a list of chord progressions with scores. More...
using	ObsId = unsigned
using	StateId = unsigned
using	Score = double

Detailed Description

template<class Obs>
class HMM_backend< Obs >

Backend for a Hidden Markov Model object, templated over the observation type.

The parameter of the template correspond to the Observation type. We can imagine several ways to describe these, for example with meta datas rather than plain music description

Member Function Documentation

template<class Obs >

std::vector<std::pair<ChordProgression,Score> > HMM_backend< Obs >::getBestProgressions ( Melody  M, unsigned  n = 1  ) const

virtual

Return a list of chord progressions with scores.

Scores are non negative. It is guaranted that at least one progression has a score different than 0.

Parameters

M	Observed melody from which we want to generate chords
n	Number of chord progressions wanted (n > 0 )

Exceptions

std::domain_error

If the number of chords requested is higher than what can be generated, an exception is thrown

Returns

A list of n chord progressions, along with their scores, in decreasing order.

Implements HMM.

template<class Obs >

unsigned HMM_backend< Obs >::getMaxProgressionsNbr ( ) const

virtual

Return the maximal number of distinct progressions that can be generated

Returns

a non negative integer, the maximal number of progressions

Implements HMM.

template<class Obs >

void HMM_backend< Obs >::saveToFile ( const std::string &  fname ) const

virtual

Save the class to a given file.

Parameters

fname

Name of the target file

Implements HMM.

---

The documentation for this class was generated from the following file:

• Markov.hpp