-
-
Notifications
You must be signed in to change notification settings - Fork 14
Synthetizer Class
This is the main module that generates the sound.
Yes, it's spelled "Synthetizer" and not "Synthesizer" here.
Tip
If you encounter any errors in this documentation, please open an issue!
- Synthetizer Class
- Table of Contents
- Importing
- Initialization
- Destruction
- Methods
- Properties
- Managers And Modes
// normal install
import { Synthetizer } from "./spessasynth_lib/synthetizer/synthetizer.js";
// npm package
import { Synthetizer } from "spessasynth_lib";
Tip
Using the npm package? Make sure you've read this
Caution
Note that you need to add the worklet processor for the synthesizer to work! See Importing the worklet
const synth = new Synthetizer(
tagetNode,
soundFontBuffer,
enableEventSystem(optional),
startRenderingData(optional),
effectsConfig(optional)
);
- targetNode - the AudioNode the synth should play to. Usually it is the
AudioContext.destination
property. - soundFontBuffer - the
ArrayBuffer
to your soundFont. - enableEventSystem -
boolean
, disables the event system. Useful when rendering audio to file - startRenderingData -
object
, used for rendering to file. It's formatted as follows:- parsedMIDI: a
MIDI
class instance. The synthesizer will immediately start rendering it if specified - snapshot: a
SynthesizerSnapshot
object, a copy of controllers from another synthesizer instance. If specified, synth will copy this configuration. - oneOutput: a
boolean
- indicates the One output mode - loopCount: the number of loops to play.
It defaults to 0.
Make sure your
OfflineAudioContext
's length accounts for these!
- parsedMIDI: a
- effectsConfig → optional, the configuration for audio effects. See below.
Caution
Avoid using multiple synthesizer instances. The SoundFontManager and one instance should be sufficient. See this comment for more info.
-
chorusConfig
-object
- this is the chorus config object:-
nodesAmount
-number
- the number of delay nodes (for each channel) and the corresponding oscillators. -
defaultDelay
-number
- the initial delay, in seconds. -
delayVariation
-number
- the difference between delays in the delay nodes. -
stereoDifference
-number
- the difference of delays between two channels (added to the left channel and subtracted from the right). -
oscillatorFrequency
-number
- the initial delay oscillator frequency, in Hz. -
oscillatorFrequencyVariation
-number
- the difference between frequencies of oscillators, in Hz. -
oscillatorGain
-number
- how much the oscillator will alter the delay in delay nodes, in seconds.
-
-
chorusEnabled
-boolean
- indicates if the chorus effect is enabled. -
chorusConfig
-ChorusConfig
- the configuration for chorus. Passundefined
to use defaults. Described above. -
reverbEnabled
-boolean
- indicates if the reverb effect is enabled. -
reverbImpulseResponse
-AudioBuffer
- the impulse response for the reverb. Passundefined
to use defaults.
Tip
Pass undefined
to chorusConfig
or reverbImpulseResponse
to use the defaults.
Important
If you're rendering the audio to a file, it is highly recommended to supply the synthesizer with the reverb buffer
(even the stock one) through effects config, because it internally fetch
es the file which will result in reverb
enabling after a second or so.
A promise that gets resolved when the worklet synthesizer gets fully initialized (including sf3 support)
await synth.isReady;
Tip
It is recommended (but not always required) to wait for this promise.
Use the .destroy()
method.
synth.destroy();
Caution
Remember, you MUST call this method after you're done with the synthesizer! Otherwise it will keep processing and the performance will greatly suffer.
Important
The synthesizer internally sends commands to the AudioWorklet
where all the processing happens.
Keep that in mind as not all methods will immediately report values!
(E.g. noteOn
won't instantly increase the voice
count in channelProperties
)
Send a raw MIDI message to the synthesizer. Calls noteOn, noteOff, etc. internally.
synth.sendMessage(message, channelOffset = 0);
- message - an array of bytes (numbers from 0 to 255). The MIDI message to process.
- channelOffset, optional - adds to the channel number of the message. It defaults to 0.
Play the given note.
synth.noteOn(channel, midiNote, velocity, enableDebugging);
- channel - the MIDI channel to use. It usually ranges from 0 to 15, but it depends on the channel count.
- midiNote - the note to play. Ranges from 0 to 127.
- velocity - controls how loud the note is.
Note that velocity of 0 has
the same effect as using
noteOff
. Ranges from 0 to 127, where 127 is the loudest and 1 is the quietest. - enableDebugging - boolean, used only for debugging.
When
true
, the console will print out tables of the soundfont generator data used to play the note.
Stop the given note.
synth.noteOff(channel, midiNote);
- channel - the MIDI channel to use. It Usually ranges from 0 to 15, but it depends on the channel count.
- midiNote - the note to play. Ranges from 0 to 127.
Note that when highPerformanceMode
is set to true, the note will always have a release time of 50ms.
Change the preset for the given channel.
synth.programChange(channel, programNumber);
- channel - the MIDI channel to change. It usually ranges from 0 to 15, but it depends on the channel count.
- programNumber - the MIDI program number to use. Ranges from 0 to 127. To use other banks, go to controllerChange.
Change the channel's pitch, including the currently playing notes.
synth.pitchWheel(channel, MSB, LSB);
- channel - the MIDI channel to use. It usually ranges from 0 to 15, but it depends on the channel count.
- MSB and LSB. 7-bit numbers that form a 14-bit pitch bend value.
Change the channel's pitch bend range in semitones. It uses Registered Parameter Number internally.
synth.setPitchBendRange(channel, pitchBendRangeSemitones);
- channel - the MIDI channel to use. It usually ranges from 0 to 15, but it depends on the channel count.
- pitchBendRangeSemitones - the pitch bend range, in full semitones.
Handle a MIDI System Exclusive message.
synth.systemExclusive(messageData);
- message data - Uint8Array, the message byte data Excluding the 0xF0 byte!
Tip
Refer to this table for the list of supported System Exclusives.
Set a given MIDI controller to a given value.
synth.controllerChange(channel, controllerNumber, controllerValue);
- channel - the MIDI channel to use. It usually ranges from 0 to 15, but it depends on the channel count.
- controllerNumber - the MIDI CC number of the controller to change. Refer to this table for the list of controllers supported by default.
- controllerValue - the value to set the given controller to. Ranges from 0 to 127.
Note
Note that theoreticallly all controllers are supported as it depends on the SoundFont's modulators.
Reset all controllers to their default values. (for every channel)
synth.resetControllers();
Cause the given midi channel to ignore controller messages for the given controller number.
synth.lockController(channel, controllerNumber, isLocked);
- channel - the channel to lock. It usually ranges from 0 to 15, but it depends on the channel count.
- controllerNumber - the MIDI CC to lock. Ranges from 0 to 146. See the tip below to see why.
- isLocked - boolean, if true then locked, if false then unlocked.
Tip
To lock other modulator sources add 128 to the Source Enumerator (Soundfont 2.04 Specification section 8.2.1)
For example to lock pitch wheel, use synth.lockController(channel, 142, true)
. (128 + 14 = 142)
Apply pressure to the given channel. It usually controls the vibrato amount.
synth.channelPressure(channel, pressure);
- channel - the channel to use. It usually ranges from 0 to 15, but it depends on the channel count.
- pressure - the pressure to apply. Ranges from 0 to 127.
Apply pressure to the given note on a given channel. It usually controls the vibrato amount.
synth.polyPressure(channel, midiNote, pressure);
- channel - the channel to use. It usually ranges from 0 to 15, but it depends on the channel count.
- midiNote - the note to apply pressure to. Ranges from 0 to 127.
- pressure - the pressure to apply. Ranges from 0 to 127.
Mute or unmute a given channel.
synth.muteChannel(channel, isMuted);
- channel - number - the channel to mute/unmute. It usually ranges from 0 to 15, but it depends on the channel count.
- isMuted - boolean - if the channel should be muted. boolean.
Force all the notes in a given channel to have the specified velocity.
synth.velocityOverride(channel, velocity);
- channel - number - the channel to use. It usually ranges from 0 to 15, but it depends on the channel count.
- velocity - number - the velocity to use. 0 disables the override.
Stop all notes. Equivalent of MIDI "panic."
synth.stopAll();
Transpose the synth up or down in semitones. Floating point values can be used for more precise tuning.
synth.transpose(semitones);
- semitones - number - the number of semitones to transpose the synth by. It can be positive or negative or zero. Zero resets the pitch.
Set the synth's main volume.
synth.setMainVolume(volume);
- volume - number - the synth's volume. Ranges from 0 to anything, but 1 is the recommended maximum.
Note
Raising the gain above 1 can lead to unexpected results.
Set the synth's interpolation method.
synth.setInterpolationType(type);
-
type - number - the interpolation type. Currently, defined types:
-
0 - linear interpolation. This was previously the default
-
1 - no interpolation (the nearest neighbor). Useful for songs like chiptunes.
-
2 - cubic (fourth) order interpolation. Default.
Add a new channel. Invokes a newchannel
event.
synth.addNewChannel();
Caution
This function is deprecated. It may be removed without further notice. Use soundfontManager instead.
Change the soundfont of a Synthesizer's instance.
await synth.reloadSoundFont(soundFontBuffer);
Important
This function is asynchronous.
- soundFont - the soundfont to change to, an
ArrayBuffer
instance of the file.
Set the new impulse response for the reverb algorhitm.
synth.setReverbResponse(buffer);
- buffer -
AudioBuffer
- contains the new impulse response to use.
Set and update the chorus processor with a given config.
synth.setChorusConfig(config);
- config -
ChorusConfig
- the new configuration. Format is described here.
Set the gain value of the built-in effects or disable them.
synth.setEffectsGain(reverbGain, chorusGain);
- reverbGain -
number
- the gain value of the reverb effect. Ranges from 0 to 1. Set to 0 to disable. - chorusGain -
number
- the gain value of the chorus effect. Ranges from 0 to 1. Set to 0 to disable.
Get a current snapshot of the Worklet synthesizer.
Important
This function is asynchronous.
const snapshot = await synth.getSynthesizerSnapshot();
The returned value is formatted like this. It is essentially an object of the entire synthesizer instance from the audioWorklet side.
Disables GS NRPN (Non-Registered Parameter Number) messages from being recognized. Such as vibrato or drum key tuning.
synth.disableGSNRPparams();
Connects individual channel outputs to given target nodes.
synth.connectIndividualOutputs(audioNodes);
- audioNodes -
audioNode[]
- an array of exactly 16AudioNodes
to connect each channel to. The first node connects to the first channel and so on.
Print out the synth class instance, both from the main thread and the AudioWorklet thread.
synth.debugMessage();
The synthesizer's event handler. Refer to Event handling for more.
The synthesizer's soundfont manager. Refer to The soundfont manager for more.
The synthesizer's key modifer manager. Refer to Key modifier manager for more.
The current amount of voices (notes) playing or during their release phase.
console.log(`This synthetizer is currently playing ${synth.voicesAmount} notes!`);
The maximum allowed voices at once. If new voices are added, the voices considered unimportand are killed. Default is 350.
synth.voiceCap = 100; // max 100 voices at once
The connected AudioContext
's time.
console.log(`The current AudioContext's time is ${synth.currentTime}!`); // example usage
Indicates the current system the synth is in. Currently, there are: GM, GM2, GS, XG. Default is GS
console.log(synth.system); // "gm"
Boolean, if the high performance mode is enabled. High performance mode currently overrides release time to be almost instant. Intended for "Black MIDIs."
synth.highPerformanceMode = true; // we can now play black MIDIs! >:)
The current channel properties. An array
of objects formatted like this:
/**
* @typedef {Object} ChannelProperty
* @property {number} voicesAmount - the channel's current voice amount
* @property {number} pitchBend - the channel's current pitch bend from -8192 do 8192
* @property {number} pitchBendRangeSemitones - the pitch bend's range, in semitones
* @property {boolean} isMuted - indicates whether the channel is muted
* @property {boolean} isDrum - indicates whether the channel is a drum channel
*/
console.log(synth.channelProperties[0]); // {voicesAmount: 0, pitchBend: 0, pitchBendRangeSemitones: 2, isMuted: false, isDrum: false }
The synthesizer supports event handling. For example, the MidiKeyboard
uses handling to visualize keypresses.
Use the property synth.eventHandler
to access the system.
synth.eventHandler.addEvent(name, id, callback);
- name - the type of the event. refer to the table below.
- id - unique id for the event listener. Can be anything, as long as it's unique.
- callback.
a function that gets called on the event.
Callback takes an
object
argument. The properties depend on the event type. Refer to the table below.
synth.eventHandler.removeEvent(name, id);
- name - the type of the event.
- id - the unique id of the event you wish to remove.
If you need to delay the events (for example, to sync up with something),
you can use the timeDelay
property.
synth.eventHandler.timeDelay = 5;
The delay time is specified in seconds. Set to 0 to disable (instant callback). Default is 0.
See this table for the list of supported events.
The soundfont manager allows for handling multiple soundfonts.
Every operation sends a new presetlist
event.
Use the soundfontManager
property.
synth.soundfontManager;
synth.soundfontManager.soundfontList;
Which is a list of objects defined as follows:
- id -
string
- the unique soundfont identifier. - bankOffset -
number
- the bank offset for the soundfont.
The list is ordered from the most important soundfont to the least (e.g., first soundfont is used as a base and other soundfonts get added on top (not override))
Important
When first creating the synthesizer,
soundfontList contains one soundfont with the identifier main
and bank offset of 0.
The behavior is defined as follows:
- The program looks for the first soundfont that has the requested program:bank combo and uses it.
- If not found, the program looks for the first soundfont that has the requested program number and uses it.
- If not found, the program uses the first preset of the first soundfont.
This function adds a new soundfont at the top of the soundfont stack.
await synth.soundfontManager.addNewSoundFont(soundfontBuffer, id, bankOffset = 0);
- soundfontBuffer -
ArrayBuffer
- the soundfont binary data. - id -
string
- unique ID for the soundfont. Any string as long as it's unique. - bankOffset -
number
, optional - the bank offset for the soundfont.
Important
This function is asynchronous.
This function removes a specified soundfont.
synth.soundfontManager.deleteSoundFont(id);
- id -
string
- unique ID for the soundfont to delete.
This function reorders the soundfonts.
synth.soundfontManager.rearrangeSoundFonts(newOrderedList);
- newOrderedList - array of
string
- The new list of the soundfont identifiers, in the desired order.
This function removes all soundfonts and adds a new one with id main
and bank offset of 0.
await synth.soundfontManager.reloadManager(soundfontBuffer);
- soundfontBuffer -
ArrayBuffer
- the new soundfont to reload the synth with.
Important
This function is asynchronous.
This powerful tool allows modifying each key on each channel to your needs.
Currently, it supports overriding:
- the velocity of that note
- the preset used on that note
This function modifies a single key.
synth.keyModifierManager.addModifier(channel, midiNote, options);
- channel - the MIDI channel to use. It usually ranges from 0 to 15, but it depends on the channel count.
- midiNote - the MIDI note to modify. Ranges from 0 to 127.
- options - the note's modifiers. An
Object
:- velocity -
number
- optional. Forces this key on this channel to be the given velocity. Unchanged if undefined. - patch -
Object
- optional. Forces this key on this channel to play with the given patch.- program -
number
- the program number of the desired patch. - bank -
number
- the bank number of the desired patch. - Note that both
program
andbank
must be provided if thepatch
option is used.
- program -
- velocity -
Clears the modifier from a note, making it behave normally.
synth.keyModifierManager.deleteModifier(channel, midiNote)
- channel - the MIDI channel to use. It usually ranges from 0 to 15, but it depends on the channel count.
- midiNote - the MIDI note to modify. Ranges from 0 to 127.
Get the key modifier for a given key on a given channel. Returns undefined
if there's none.
synth.keyModifierManager.getModifier(channel, midiNote)
- channel - the MIDI channel to use. It usually ranges from 0 to 15, but it depends on the channel count.
- midiNote - the MIDI note to modify. Ranges from 0 to 127.
The returned value is a KeyModifier
object.
Clears ALL modifiers in this synthesizer instance.
synth.keyModifierManager.clearModifiers();
This is a special synth mode, which causes the synth to have one output (instead of 18), but 32 channels.
Every midi channel has two audio channels. So it looks like this:
- MIDI channel 0:
- audio output 0
- audio output 1
- MIDI channel 1:
- audio output 2
- audio output 3
- MIDI channel 2:
- audio output 4
- audio output 5
etc.
This allows for many things, such as exporting files of individual channels.
const midi = YOUR_MIDI_HERE;
const soundfont = YOUR_SOUNDFONT_HERE;
const sampleRate = 44100;
const offline = new OfflineAudioContext({
numberOfChannels: 32,
length: sampleRate * midi.duration,
sampleRate: sampleRate
});
const synth = new Synthetizer(
offline.destination,
soundfont,
false, // disable event system for faster rendering
{
parsedMIDI: midi,
oneOutput: true
}
);
// render
const renderedData = await offline.startRendering();
for (let i = 0; i < 16; i++)
{
const audioOut = audioBufferToWav(buf, false, i * 2);
// do whatever you need with the channel buffer
console.log('Rendered channel', i, 'data:', audioOut);
}
Important
Chorus and reverb are disabled when the one output mode is on.
Caution
The OfflineAudioContext must be initialized with 32 channels! Otherwise there will be an error!
Tip
If you encounter any errors in this documentation, please open an issue!
Warning
Make sure you always update worklet_processor.min.js
along with the npm package!