annotate vamp-sdk/Plugin.h @ 3:0133b3513e2b

* Renamed sdk to vamp-sdk

/* -*- c-basic-offset: 4 indent-tabs-mode: nil -*-  vi:set ts=8 sts=4 sw=4: */

/*
    Vamp

    An API for audio analysis and feature extraction plugins.

    Centre for Digital Music, Queen Mary, University of London.
    Copyright 2006 Chris Cannam.
  
    Permission is hereby granted, free of charge, to any person
    obtaining a copy of this software and associated documentation
    files (the "Software"), to deal in the Software without
    restriction, including without limitation the rights to use, copy,
    modify, merge, publish, distribute, sublicense, and/or sell copies
    of the Software, and to permit persons to whom the Software is
    furnished to do so, subject to the following conditions:

    The above copyright notice and this permission notice shall be
    included in all copies or substantial portions of the Software.

    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
    EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
    MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
    NONINFRINGEMENT. IN NO EVENT SHALL THE X CONSORTIUM BE LIABLE FOR
    ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
    CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
    WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

    Except as contained in this notice, the names of the Centre for
    Digital Music; Queen Mary, University of London; and Chris Cannam
    shall not be used in advertising or otherwise to promote the sale,
    use or other dealings in this Software without prior written
    authorization.
*/

#ifndef _VAMP_PLUGIN_H_
#define _VAMP_PLUGIN_H_

#include "PluginBase.h"
#include "RealTime.h"

#include <string>
#include <vector>
#include <map>

namespace Vamp {

/**
 * Vamp::Plugin is a base class for plugin instance classes
 * that provide feature extraction from audio or related data.
 *
 * In most cases, the input will be audio and the output will be a
 * stream of derived data at a lower sampling resolution than the
 * input.
 *
 * Note that this class inherits several abstract methods from
 * PluginBase, that must be implemented by the subclass.
 */

/**
 * Plugin Lifecycle
 * ================
 *
 * Feature extraction plugins are managed differently from real-time
 * plugins (such as VST effects).  The main difference is that the
 * parameters for a feature extraction plugin are configured before
 * the plugin is used, and do not change during use.
 *
 * 1. Host constructs the plugin, passing it the input sample rate.
 * The plugin may do basic initialisation, but should not do anything
 * computationally expensive at this point.
 *
 * 2. Host may query the plugin's available outputs.
 *
 * 3. Host queries programs and parameter descriptors, and may set
 * some or all of them.  Parameters that are not explicitly set should
 * take their default values as specified in the parameter descriptor.
 * When a program is set, the parameter values may change and the host
 * will re-query them to check.
 *
 * 4. Host queries the preferred step size, block size, number of
 * channels, and the number of values per feature for the plugin's
 * outputs.  These may all vary depending on the parameter values.
 * (Note however that you cannot make the number of distinct outputs
 * dependent on parameter values; nor can you make any of these depend
 * on the number of input channels.)
 *
 * 5. Plugin is properly initialised with a call to initialise.  This
 * fixes the step size, block size, and number of channels, as well as
 * all of the parameter and program settings.  If the values passed in
 * to initialise do not match the plugin's advertised preferred values
 * from step 4, the plugin may refuse to initialise and return false
 * (although if possible it should accept the new values).
 *
 * 6. Host will repeatedly call the process method to pass in blocks
 * of input data.  This method may return features extracted from that
 * data (if the plugin is causal).
 *
 * 7. Host will call getRemainingFeatures exactly once, after all the
 * input data has been processed.  This may return any non-causal or
 * leftover features.
 *
 * 8. At any point after initialise was called, the host may
 * optionally call the reset method and restart processing.  (This
 * does not mean it can change the parameters, which are fixed from
 * initialise until destruction.)
 *
 * A plugin does not need to handle the case where setParameter or
 * selectProgram is called after initialise has been called.  It's the
 * host's responsibility not to do that.
 */

class Plugin : public PluginBase
{
public:
    /**
     * Initialise a plugin to prepare it for use with the given number
     * of input channels, step size (window increment, in sample
     * frames) and block size (window size, in sample frames).
     *
     * The input sample rate should have been already specified at
     * construction time.
     * 
     * Return true for successful initialisation, false if the number
     * of input channels, step size and/or block size cannot be
     * supported.
     */
    virtual bool initialise(size_t inputChannels,
			    size_t stepSize,
			    size_t blockSize) = 0;

    /**
     * Reset the plugin after use, to prepare it for another clean
     * run.  Not called for the first initialisation (i.e. initialise
     * must also do a reset).
     */
    virtual void reset() = 0;

    enum InputDomain { TimeDomain, FrequencyDomain };
    
    /**
     * Get the plugin's required input domain.  If this is TimeDomain,
     * the samples provided to the process() function (below) will be
     * in the time domain, as for a traditional audio processing
     * plugin.  If this is FrequencyDomain, the host will carry out a
     * windowed FFT of size equal to the negotiated block size on the
     * data before passing the frequency bin data in to process().
     * The plugin does not get to choose the window type -- the host
     * will either let the user do so, or will use a Hanning window.
     */
    virtual InputDomain getInputDomain() const = 0;

    /**
     * Get the preferred step size (window increment -- the distance
     * in sample frames between the start frames of consecutive blocks
     * passed to the process() function) for the plugin.  This should
     * be called before initialise().
     */
    virtual size_t getPreferredStepSize() const = 0;

    /**
     * Get the preferred block size (window size -- the number of
     * sample frames passed in each block to the process() function).
     * This should be called before initialise().
     */
    virtual size_t getPreferredBlockSize() const { return getPreferredStepSize(); }

    /**
     * Get the minimum supported number of input channels.
     */
    virtual size_t getMinChannelCount() const { return 1; }

    /**
     * Get the maximum supported number of input channels.
     */
    virtual size_t getMaxChannelCount() const { return 1; }

    struct OutputDescriptor
    {
	/**
	 * The name of the output, in computer-usable form.  Should be
	 * reasonably short and without whitespace or punctuation.
	 */
	std::string name;

	/**
	 * The human-readable name of the output.
	 */
	std::string description;

	/**
	 * The unit of the output, in human-readable form.
	 */
	std::string unit;

	/**
	 * True if the output has the same number of values per result
	 * for every output result.  Outputs for which this is false
	 * are unlikely to be very useful in a general-purpose host.
	 */
	bool hasFixedValueCount;

	/**
	 * The number of values per result of the output.  Undefined
	 * if hasFixedValueCount is false.  If this is zero, the output
	 * is point data (i.e. only the time of each output is of
	 * interest, the value list will be empty).
	 *
	 * Note that this gives the number of values of a single
	 * output result, not of the output stream (which has one more
	 * dimension: time).
	 */
	size_t valueCount;

	/**
	 * The names of each of the values, if appropriate.  This is
	 * always optional.
	 */
	std::vector<std::string> valueNames;

	/**
	 * True if the results in the output have a fixed numeric
	 * range (minimum and maximum values).  Undefined if
	 * valueCount is zero.
	 */
	bool hasKnownExtents;

	/**
	 * Minimum value of the results in the output.  Undefined if
	 * hasKnownExtents is false or valueCount is zero.
	 */
	float minValue;

	/**
	 * Maximum value of the results in the output.  Undefined if
	 * hasKnownExtents is false or valueCount is zero.
	 */
	float maxValue;

	/**
	 * True if the output values are quantized to a particular
	 * resolution.  Undefined if valueCount is zero.
	 */
	bool isQuantized;

	/**
	 * Quantization resolution of the output values (e.g. 1.0 if
	 * they are all integers).  Undefined if isQuantized is false
	 * or valueCount is zero.
	 */
	float quantizeStep;

	enum SampleType {

	    /// Results from each process() align with that call's block start
	    OneSamplePerStep,

	    /// Results are evenly spaced in time (sampleRate specified below)
	    FixedSampleRate,

	    /// Results are unevenly spaced and have individual timestamps
	    VariableSampleRate
	};

	/**
	 * Positioning in time of the output results.
	 */
	SampleType sampleType;

	/**
	 * Sample rate of the output results.  Undefined if sampleType
	 * is OneSamplePerStep.
	 *
	 * If sampleType is VariableSampleRate and this value is
	 * non-zero, then it may be used to calculate a resolution for
	 * the output (i.e. the "duration" of each value, in time).
	 * It's recommended to set this to zero if that behaviour is
	 * not desired.
	 */
	float sampleRate;
    };

    typedef std::vector<OutputDescriptor> OutputList;

    /**
     * Get the outputs of this plugin.  An output's index in this list
     * is used as its numeric index when looking it up in the
     * FeatureSet returned from the process() call.
     */
    virtual OutputList getOutputDescriptors() const = 0;

    struct Feature
    {
	/**
	 * True if an output feature has its own timestamp.  This is
	 * mandatory if the output has VariableSampleRate, and is
	 * likely to be disregarded otherwise.
	 */
	bool hasTimestamp;

	/**
	 * Timestamp of the output feature.  This is mandatory if the
	 * output has VariableSampleRate, and is likely to be
	 * disregarded otherwise.  Undefined if hasTimestamp is false.
	 */
	RealTime timestamp;
	
	/**
	 * Results for a single sample of this feature.  If the output
	 * hasFixedValueCount, there must be the same number of values
	 * as the output's valueCount count.
	 */
	std::vector<float> values;

	/**
	 * Label for the sample of this feature.
	 */
	std::string label;
    };

    typedef std::vector<Feature> FeatureList;
    typedef std::map<int, FeatureList> FeatureSet; // key is output no

    /**
     * Process a single block of input data.
     * 
     * If the plugin's inputDomain is TimeDomain, inputBuffers will
     * point to one array of floats per input channel, and each of
     * these arrays will contain blockSize consecutive audio samples
     * (the host will zero-pad as necessary).
     *
     * If the plugin's inputDomain is FrequencyDomain, inputBuffers
     * will point to one array of floats per input channel, and each
     * of these arrays will contain blockSize/2 consecutive pairs of
     * real and imaginary component floats corresponding to bins
     * 0..(blockSize/2-1) of the FFT output.
     *
     * The timestamp is the real time in seconds of the start of the
     * supplied block of samples.
     *
     * Return any features that have become available after this
     * process call.  (These do not necessarily have to fall within
     * the process block, except for OneSamplePerStep outputs.)
     */
    virtual FeatureSet process(float **inputBuffers,
			       RealTime timestamp) = 0;

    /**
     * After all blocks have been processed, calculate and return any
     * remaining features derived from the complete input.
     */
    virtual FeatureSet getRemainingFeatures() = 0;

    virtual std::string getType() const { return "Feature Extraction Plugin"; }

protected:
    Plugin(float inputSampleRate) :
	m_inputSampleRate(inputSampleRate) { }

    float m_inputSampleRate;
};

}

#endif