drumclone

Drumclone is in development stage and currently provides a single model tuned for kick-drums (drumclone kick-modelling beta).

Drumclone kick-modelling is free to use:

⚠ warning: drumclone will be in beta until proven to be stable. Consider → reporting any errors you encounter.

* This version also works for other types of drum sounds, but will require some manual input and experimentation.

* Future models might require a paid license to support their development and maintainance. (e.g. a one-time purchase).

* Check →system requirements and →installation for setup instructions.

Drumclone allows extracting and modifying high-quality single drum sounds from complex audio sources like songs, samples, or any recording with perceptible percussive elements.

Runs as a VST, to be used in DAWs or any kind of host with VST3 support.

Drumclone comes with models to precisely isolate and resynthesize drum sounds, which allows a fast iterative workflow thanks to its sub-second processing times and drag and drop UI.

While it works for isolation, drumclone’s main focus is it’s sound-design capabilities. Being able to imitate one-shot percussive noises and tweak their attack, transient, noise component, etc, down to the sample-level.

The utilization of pre-made sounds or song samples is a prevalent practice across various music production genres. Drum samples, in particular, serve as essential components, often forming the structural foundation of numerous compositions. Utilizing drumclone, these high-quality drum sounds can be extracted from existing recordings and subsequently modified without any degradation in sound quality.

This methodology can effectively isolate a drum sound and mutate it from pure sine-like percussion, to harmonically rich and distorted drum sounds, while keeping the processing fast enough to visualize the results it in real-time.

Unlike traditional extraction methods, which tend to degrade the quality of the sounds with unnatural filtering artifacts and output static audio samples, resynthesis can model the original timbral characteristics of the drums and allows creative flexibility by tuning the underlying formula with simple user-facing controls.

1. One-shot percussion isolation and resynthesis.
2. Flexible transient and body sculpting with tone and noise controls.
3. Fine-tuning with sample-level resolution.
4. Synthesis models to avoid typical filtering isolation artifacts.
5. Extremely clear drum sound output for applying further processing chains.
6. Real-time visualization of every change in the waveform.
7. Fast drag-and-drop workflow.
8. Fully offline, no cloud or servers involved.
9. Sub-second processing times on common hardware.
10. Multi-segment envelope editors to accurately shape the underlying model.
11. Low-profile vector UI with data visualization and inline controls.

The following list of drumclone outputs shows the clarity and flexibility of the current kick drum model in multiple contexts.

Processing starts with time and frequency domain analysis of the input. Features are extracted and modeled in the time domain to aid in detecting events and components corresponding to percussive signals in the spectral domain.

Components are isolated from the input using different soft spectral masks tailored to the component type, built around detected events and user-specified envelopes.

Once these audio layers are reconstructed from the spectral domain, dynamics and stereo processing are applied to refine the output.

These filtered percussive component layers serve as the input for the resynthesis process. Each signal is matched using a synthesis technique that accurately represents their tonal and amplitude characteristics.

This is achieved by using different polyphonic or monophonic oscillators (sine and band-passed noise) and generating modulation envelopes that closely resemble the input.

Following resynthesis, the layers are mixed down into a cohesive stereo track. Straightforward cross-fading and summing with predefined curves are employed in the mix.

Finally, stereo and dynamics processing are applied to the output track to achieve a more natural sound and controlled audio levels.