Latency

Real-time detection latency is determined primarily by the internal processing pipeline: resampling, filtering and the analysis window. This fixed component cannot be reduced. The user-adjustable factors are the audio buffer size and the onset detection threshold.

Buffer size — determines how much audio is processed at once. Smaller buffers reduce latency but increase the risk of underruns. For minimum latency, set the buffer size to between 32 and 128 samples and the sample rate to 48 kHz.

Onset detection threshold — controlled by the Sensitivity parameter. A higher value lowers the threshold, reducing the delay between a true onset and its detection — specifically, it reduces the rise time of the modelled transient response. For live performance, set Sensitivity to 0.9. This minimises detection delay at the cost of increased responsiveness to unintended transients.

File conversion does not operate under real-time constraints. Latency settings have no effect on offline conversion. For batch processing, the standalone application is often more practical — it runs independently, without a DAW or processing chain.