SketchSynth Dataset

Dataset of sound-sketches produced by 62 participants to the semantic prompt pairs: noisy - calm, thin - thick, clean - rough The JSON file containing the raw sketch and is formatted as follows: <pre><code>{ participant1_id: { prompt1: [...], prompt2: [...], prompt3: [...], ...}, participant2_id: { prompt1: [...], prompt2: [...], prompt3: [...], ...}, . . . }</code></pre> Sketch data is saved in the format: <pre><code>#Sketch array [#Stroke 1 array [#Recorded stroke points [x0,x1,x2,...], #x positions [y0,y1,y2,...], #y positions [t0,t1,t2,...], #timestamps ], #More Stroke arrays [...], [...], ... ]</code></pre> The dataset also includes rasterised sketches for each participant and each prompt, the predicted sounds and the ratings that participant provided.