Comparing Synthesizers
spessasynth_lib provides two synthesizers:
This synthesizer uses a single AudioWorklet to provide real-time playback.
Advantages
- Stable audio system: The processor lives in the worklet. As long as the system can keep up, there should be no dropouts.
- Separate thread: The audio thread is separated from the main thread, ensuring smooth playback even if the main thread is busy.
- Fast main thread communication: Suitable for real-time playback from Web MIDI inputs.
- Simple to set up.
Disadvantages
- Susceptible to the Chromium SF3 audio bug with no workarounds. Weak devices with Chromium can be unusable.
- Rendering audio requires another instance of the synthesizer. This potentially means copying a 4GB sound bank buffer twice.
- No access to the audio engine or ability to intercept the messages.
This newer synthesizer uses a Worker communicating with an AudioWorklet through a MessageChannel to provide real-time playback.
Advantages
- Separate thread: The worker thread is separated from the main thread, ensuring smooth playback even if the main thread is busy.
- Integrated worklet: There's no need to copy any JS files as the worklet is embedded into the code.
- Direct audio engine access: While still in a separate thread.
- Ability to intercept messages passed from and to the core synthesis engine.
- Not affected by the Chromium SF3 audio bug.
- Direct SF/MIDI/audio export right in the worker, without transferring any large ArrayBuffers or freezing the main thread.
Disadvantages
- Less stable audio engine: The audio processing is separated into two threads, increasing the potential for stutters.
- Less accurate real-time inputs: Calling note-on, note-off, etc. in real-time is less precise than with the Worklet.
- A lot of clicks and stutters on non-Chromium browsers on weaker devices such as smartphones.
- More code needed to set it up.