k2k audio Parameters
[Parameters not yet documented — likely declared in source outside register_animatable(). Add an override in _overrides/ if needed.]
Additional controls
File Path — The audio file feeding the graph. Drag a file directly onto the drop zone, or click Browse… to open the audio browser (which lets you preview files before loading). Supported formats are WAV, AIFF, and FLAC. The path is stored in the saved graph, so loading the graph again on the same machine restores the source automatically — but file paths don’t travel between machines, so when sharing graphs the recipient will need to relink the source.
Normalize — When on, scales the loaded audio so its loudest peak hits 0 dBFS — useful for evening out levels across a session of varied source material. Off keeps the file at its original level. Normalization happens once at load time, not as live processing, so toggling this re-runs the load. For loud or commercial-master sources you generally want this off; for quiet field recordings or stems, on.
Fft Size — Window size for the STFT (short-time Fourier transform), 512–8192 samples. This sets the time/frequency tradeoff for everything downstream in the graph. Larger windows (4096, 8192) give finer frequency resolution — better for low-pitched material, sustained tones, and tonal/spectral work; downside is smearing of fast events in time. Smaller windows (512, 1024) give finer time resolution — better for transients and percussive material; downside is coarser pitch and frequency detail. 4096 is a balanced default for most material; drop to 1024 for drum-heavy sources, push to 8192 for slow drones or pitch-focused processing. Changing this here affects every spectral-domain node downstream.
Hop Size — Number of samples the STFT analysis window advances between frames, 128 to fft_size/2. The ratio of hop_size to fft_size sets the overlap percentage (shown live below the field). 25% hop (75% overlap) is the standard default — high enough overlap for clean reconstruction without excessive CPU. Lower hop sizes (more overlap) give smoother time interpolation at the cost of more frames to process; higher hop sizes (less overlap) are faster but can introduce frame-edge artifacts on pitch-shifting and time-stretching nodes. If you don’t know what to set, leave it at the default — fft_size/4 is set automatically when you change fft_size.
About Load Audio File
LoadAudio is the source for almost every K2K graph. The node performs the STFT once at load time, producing a ComplexBuffer that holds both the audio and its frequency-domain representation as lazy views — downstream nodes pull whichever view they need without re-analysis. Stereo files automatically expose two outputs (Left, Right); mono files have a single output. The FFT settings here become the global STFT frame for the graph, which is why most workflows configure them at the source rather than mid-chain. If you need to switch FFT parameters mid-graph, that requires re-analysis (handled by some downstream nodes) and is generally avoided when possible.
Generated 2026-05-05 from K2K_Dev@96730bdc by scripts/gen_lexique.py. Edit _intros/ or _overrides/, not this file.