VOSIMSynth is a free, fully modular audio synthesizer with a focus on speed, audio quality, and ease of use
- Speed. The real-time audio code includes very few dynamic memory allocations (including usage of STL containers), and no dynamic memory allocations in the hot path. Communication with the GUI even uses lock-free queues to minimize the chance of latency spikes. Benchmarking code is included for many of the real-time audio components.
- Audio Quality. Band-limited waveforms are generated during compilation and are resampled on the fly using massively oversampled sinc kernels to preserve fidelity at different pitches. To keep things speedy, the waveforms are first resampled to each octave during initialization. When playing a note, the waveform closest in size is used for further resampling to minimize convolution length.
- Flexibility. Any output may be connected to any input. Even feedback loops are permissible as long as a the unit delay is inserted somewhere in the loop.
- Extensibility. New units can be easily created by deriving from the Unit class. Macros are used to achieve a certain degree of reflection to automate things like voice cloning and serialization.