Performance: Accuracy and CPU Load

Timing accuracy

  • MIDI Sequencer and Audio Synthesizer run in separate system threads, not in the Unity thread, consequently playing music has no impact on your game or application timing.
  • You can modify Unity Audio Parameters from the MPTK prefab. So, synth rate and buffer size could be adapted to your need.
    • lower buffer size to lower the latency.
    • higher rate rise the CPU load and therefor lower the timing accuracy.

Tip: avoid using Unity Update() and Unity Coroutine to create rythmic music. They are not stable enough for creating music. Rather use, system thread or system event. The best approach is to use the Maestro (pro) call back method: OnAudioFrameStart. Look here.

Synth Parameters – Impact on DSP/CPU Performance

The main synthesizer parameters which impact performance are:

  • Interpolation: it’s the process of using two or more data samples to create a data sample between them.
  • Rate: at which rate the Unity internal synth (DSP) is playing the sample.
  • Buffer size: length of the sample buffer. Its is build at each cycle of the DSP.

The measure below has been done with the demo TestMidiStream. With this configuration, 25 voices are playing simultaneously:

  • Piano with one sample
  • New note every 0.2 second 
  • Duration of 5 seconds

Effect Parameters – Impact on Performance

Effects are defined in the SoundFont for each instrument. So, each of them can have a dedicated set of effects. See here. Conversely, Unity effect apply to the whole AudioSource. So, same effect is applied to all the instruments in the Midi.

Remark:

  • Effects are available only with the Pro version of MPTK.
  • By default, effects are disabled in MPTK prefab. You have to enable them from the dedicated inspector or by script with the MPTK API.

The measure has been done only for the SoundFont effect.

First test:

  • Reed organ with two samples
  • One continuous note

Second test:

  • Piano with one sample or reed organ with two samples
  • New note every 0.2 second 
  • Duration of 5 seconds

Two others possibilities: play with sample duration

  1. Increase the cut off volume and lower the device performance:

2.  Lowering the default soundfont release time: the full duration of all samples will be decreased.

The release time is the last phase of the ADSR process: time taken for the level to decay from the sustain level to zero after the key is released.

Info: this setting will be moved in the foldout synth with the next version.