3 ms·
Photosounder.com - Image-sound editor & synthesizer
- anigbrowl 17y agoI'm a fan of additive synthesis (the technology in use here) and recommend this to anyone interested in electronic music/sound design, with two caveats: a) you are likely to get the most sonically interesting results by taking recordings of sounds you already like and converting them to image files, then applying a variety of filters and treatments in photoshop or similar, before exporting back to Photosounder for rendition as audio...although this is necessarily a rather slow process. If you like Javascript, you could get into Processing, and achieve the same effect by sweeping a windowed fast-fourier transform across your audio sample (or doing a reverse FFT on vertical strips of an image). The weird digital quality of the sounds generated by a tool like this stem from harmonic juxtapositions that do not exist in nature (see below). b) the big problem with this and all other spectrogram tools I've used over the years if that they treat the audible frequency spectrum as a flat geometrical (edit: cartesian, sorry) space. So if you look at a spectrogram of a sine wave, you'll see a single horizontal line; if you examine a saw wave (which includes all harmonics) you'll see a group of horizontal lines of decreasing intensity from bottom to top. So far so good, because we can learn a lot from that, but an FFT/Spectrogram obscures the concept of pitch. For example, in conventional western tuning a 440 Hz tone is the note A, while tones of 220 or 880 Hz are the same note played an octave lower and higher, respectively. Our brains are 'wired' to pay particular attention to powers of 2; not just in pure tones, but even in distorted ones. People generally express a liking for the sound of second-harmonic distortion (at double the fundamental, the typical electric guitar sound) while exhibiting an intense dislike for third-harmonic distortion (bad transistor radio). My point here is that the sound spectrum, unlike that of light, encompasses multiple octaves and psychoacoustic research provides strong evidence for the idea that the brain evaluates sounds with multiple frequency components around a pitch spiral rather than a 2d cartesian space...but although we have the required computer power, there has been little progress in the development of tools to represent and manipulate sound this way. It's hard to find any kind of real-time tools that map the results of an FFT onto a polar coordinate system. EDIT: there's some information about construction of the polar FFT here, but it's regrettably beyond my capacity to implement at present: http://www.math.umn.edu/~focm/c_/Elad.pdf http://www.math.umn.edu/~focm/c_/Elad.pdf