Michel Rouzic - Photosounder (image-sound editing program)
greede
Started Topics :
9
Posts :
82
Posted : Mar 24, 2010 01:30:18
I just found this, anyone used it before?
Photosounder is a one-of-a-kind image-sound editing program. It is unique in that it opens images and sounds indiscriminately, treats and processes them as images, and synthesizes them as sounds. Sounds, once turned into images, can be powerfully modified to achieve effects and results that couldn't be obtained in any other way, while images of all sorts reveal the infinite kinds of otherworldly sounds they contain. Ultimately, knowing how sounds look and how images sound, you'll be able to create images that sound like what you want to hear, or like what you couldn't imagine to hear.
Photosounder is a cutting edge spectral editing program, offering the best spectrogram editing and synthesis capabilities, making use of unique spectrograph, synthesis and filtering algorithms developed specifically to achieve the best results possible. Photosounder turns sound processing problems into image processing challenges, and brings the power and flexibility of familiar image processing tools to the creation and transformation of sounds. This groundbreaking approach is what allows us to push the boundary of what we thought was possible.
Psydust
IsraTrance Junior Member
Started Topics :
14
Posts :
91
Posted : Mar 26, 2010 22:13
pretty cool prog to show to your girl friends.
and say "This is what you sound like honey"
Maine Coon
IsraTrance Junior Member
Started Topics :
12
Posts :
1659
Posted : Mar 26, 2010 23:30
Quote:
I guess they just play a sine wave kinda sound according to the pixel position?
That’s exactly what they do.
The program treats an image as a spectrogram. Each column of pixels corresponds to a single frame of sound in time. Each row corresponds to a specific frequency or a frequency band. So, every pixel tells you how loud a particular frequency is at a particular moment – the sound intensity corresponds to the pixel’s brightness.
So, when there are bright lines in a picture, going like this / , you’ll hear an upward sliding sound. When they go like this \ , the sound slides downward. This ~ will sound like vibrato. And this H will sound like “soos” because vertical lines become short bursts of white noise and the horizontal line becomes a midrange sine wave note. Little bright spots will sound like stabs. Short horizontal stripes near the bottom of the picture will sound like dirty bass or like a kick.
There are two very different ways, in which the program converts images and sounds into each other. One way is to assign exact frequencies to each row. Let’s say you tell the program that you want to see only the 32-512 Hz range, in the log scale, 1 pixel per octave. It will analyze your sound, calculate how loud the 32, 64, 128, 256 and 512 Hz components are and plot them. You tell the program to have 12 pixels per octave – it will space them by semi-tones. But you are missing everything in between those clean-cut frequencies.
So, the other way is to assign the whole frequency band to each pixel row. In our example with 12 pixels per octave it will be all the sound in between two semitone notes. It may be less precise this way but this way you are not missing anything. So, if you have a hi-hat or a crash, they will “sparkle” with their noisy spectrum instead of sounding like somebody sat on a piano. On the other hand, clear pitched instruments will sound a bit noisy, because instead of just reproducing exact harmonics the program will add frequencies that were not there in the original sound.
It’s not a big deal, though. If you specify fine enough frequency resolution (let’s say 10 cents per row, which is 120 rows per octave), you’ll get both pitched and noisy sounds reproduced quite well.
In this “noise mode” this program works exactly like a vocoder. Each row is like a band in the modulator sound. And white noise is the carrier. So, when they took an image of a synth sound and an image of Hal’s voice, layered it in Photoshop and selected “Multiply” in the layer options – they also emulated a vocoder. So, what you saw in that video example would look like this in your DAW: two chained vocoders, one with Hal’s voice as a modulator and a synth as a carrier (the Photoshop trick), the other one with the first vocoder’s output (picture) as a modulator and white noise as a carrier (the program reading the picture).
I once stretched my little niece’s singing recording in time, something like 50-fold. It created these cool slowly evolving space pads from each syllable
piko_bianko
Oxya
Started Topics :
57
Posts :
974
Posted : Mar 27, 2010 05:04
