#include <lv2synth.hpp>
Public Member Functions | |
Synth (uint32_t ports, uint32_t midi_input) | |
~Synth () | |
unsigned | find_free_voice (unsigned char key, unsigned char velocity) |
void | handle_midi (uint32_t size, unsigned char *data) |
void | pre_process (uint32_t from, uint32_t to) |
void | post_process (uint32_t from, uint32_t to) |
void | run (uint32_t sample_count) |
void | add_audio_outputs (uint32_t p1=-1, uint32_t p2=-1, uint32_t p3=-1, uint32_t p4=-1, uint32_t p5=-1, uint32_t p6=-1) |
void | add_voices (V *v01=0, V *v02=0, V *v03=0, V *v04=0, V *v05=0, V *v06=0, V *v07=0, V *v08=0, V *v09=0, V *v10=0, V *v11=0, V *v12=0, V *v13=0, V *v14=0, V *v15=0, V *v16=0, V *v17=0, V *v18=0, V *v19=0, V *v20=0) |
![]() | |
Plugin (uint32_t ports) | |
void | connect_port (uint32_t port, void *data_location) |
void | activate () |
void | deactivate () |
![]() | |
bool | check_ok () |
Protected Member Functions | |
template<typename T > | |
T *& | p (uint32_t port) |
float *& | p (uint32_t port) |
![]() | |
const char * | bundle_path () const |
void | set_ok (bool ok) |
Additional Inherited Members | |
![]() | |
static unsigned | register_class (const std::string &uri) |
This is a base class template for LV2 synth plugins. Its parameters are a voice class, either a subclass of Voice or something written from scratch, and the derived class itself. By using the derived class as a template parameter the base class can call member functions in the derived class without resorting to virtual function calls, which are hard to optimise. You can use @ref pluginmixins "mixins" with this class just like with the Plugin class, but don't use URIMap - it is already added automatically. Here is an example of a complete synth plugin. Granted, not a very exciting one, but it should be enough to explain how to do it: @code
#include <cstdlib> #include <lv2synth.hpp>
enum { MIDI_PORT, AUDIO_PORT, NUM_PORTS };
struct NoiseVoice : public LV2::Voice {
NoiseVoice() : m_gain(0), m_key(INVALID_KEY) { }
void on(unsigned char k, unsigned char velocity) { m_gain = 1.0; m_key = k; }
void off(unsigned char velocity) { m_gain = 0.0; m_key = LV2::INVALID_KEY; }
unsigned char get_key() const { return m_key; }
void render(uint32_t from, uint32_t to) { for (uint32_t i = from; i < to; ++i) { p(AUDIO_PORT)[i] += m_gain * 2 * (rand() / float(RAND_MAX) - 0.5); m_gain *= 0.9999; } }
float m_gain; unsigned char m_key; };
struct NoiseSynth : public LV2::Synth<NoiseVoice, NoiseSynth> {
NoiseSynth(double) : LV2::Synth<NoiseVoice, NoiseSynth>(NUM_PORTS, MIDI_PORT), m_filterstate(0) { add_voices(new NoiseVoice, new NoiseVoice, new NoiseVoice); add_audio_outputs(AUDIO_PORT); }
void post_process(uint32_t from, uint32_t to) { for (uint32_t i = from; i < to; ++i) { p(AUDIO_PORT)[i] = m_filterstate * 0.9 + p(AUDIO_PORT)[i] * 0.1; m_filterstate = p(AUDIO_PORT)[i]; } }
float m_filterstate; };
|
inline |
This constructor must be called in the initialiser list for your synth class.
ports | The total number of ports in your plugin. |
midi_input | The index of the main MIDI input port (the one that the synth should use for note input). |
|
inline |
This is needed to delete the voices.
|
inline |
This should be called in the constructor of your derived class to tell the synth which ports the voices are using as audio output buffers, so the synth engine can fill them with zeros before rendering the voice audio.
If you for some reason don't want to zero your audio output buffers before the voices get mixed (you may for example want to fill them with something else in pre_process()) you should not call this function - but then you MUST take care of initialising the buffers yourself in every cycle.
|
inline |
This is typically called in the constructor of your derived class to add voices to the synth. The number of voices you add determine the polyphony. It can also be called while the plugin is running to change the polyphony on the fly, but it is NOT realtime safe and NOT threadsafe - you have to make sure that the run() callback isn't executing simultaneously with this function.
LV2::Synth will assume ownership of the voices and delete them in its destructor.
|
inline |
This function implements the voice stealing algorithm. The key
and velocity
arguments are the parameters for the MIDI Note On event that needs a voice. This default implementation just returns the first free voice, and if there is none it steals voice 0.
This is not a virtual function, but if you override it in a subclass this class will still use that implementation thanks to the second template class parameter. This means that you can override this function if you want to implement your own voice stealing algorithm.
Referenced by LV2::Synth< V, D, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >::handle_midi().
|
inline |
This function is called for every MIDI event in the input. It will respond to note on and note off events by calling on() and off() in the voices.
This is not a virtual function, but if you override it in a subclass this class will still use that implementation thanks to the second template class parameter. This means that you can override this function if you want to respond to other MIDI events - just be sure to either call this function for note on and note off events or turn on and off the voices yourself.
References LV2::Synth< V, D, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >::find_free_voice().
Referenced by LV2::Synth< V, D, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >::run().
|
inlineprotected |
Use this function to access data buffers for ports. They will be casted to pointers to the template parameter T
.
Reimplemented from LV2::Plugin< D, URIMap< true >, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >.
Referenced by LV2::Synth< V, D, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >::run().
|
inlineprotected |
Use this function to access data buffers for control or audio ports.
Reimplemented from LV2::Plugin< D, URIMap< true >, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >.
|
inline |
This function is called after the synth renders a chunk of audio from the voices, from sample from
to sample to
in the output buffers. It can be used to apply global effects to the mixed audio. This default implementation does nothing.
This is not a virtual function, but if you override it in a subclass this class will still use that implementation thanks to the second template class parameter. This means that you can override this function if you want to implement your own post-processing.
Referenced by LV2::Synth< V, D, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >::run().
|
inline |
This function is called before the synth renders a chunk of audio from the voices, from sample from
to sample to
in the output buffers. It can be used to compute parameters for the voices or adding initial ambient noise to the output buffers or something else. This default implementation does nothing.
This is not a virtual function, but if you override it in a subclass this class will still use that implementation thanks to the second template class parameter. This means that you can override this function if you want to implement your own pre-processing.
Referenced by LV2::Synth< V, D, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >::run().
|
inline |
This is the main run function. It handles incoming MIDI events and mixes the voices to the output buffers with pre and post-processing applied. Don't override it unless you really know what you're doing.
Reimplemented from LV2::Plugin< D, URIMap< true >, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >.
References LV2::Synth< V, D, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >::handle_midi(), LV2::Synth< V, D, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >::p(), LV2::Synth< V, D, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >::post_process(), and LV2::Synth< V, D, Ext1, Ext2, Ext3, Ext4, Ext5, Ext6, Ext7 >::pre_process().