Michel Rouzic has just released version 1.4 of Photosounder that includes a new “lossless” mode so the output is identical to the input. Previously there was some loss of resolution importing the audio. From Michel:

Basically the lossless mode in question is a sort of 2D time-frequency filtering mode, kind of like some other programs like Audition 3 do by letting you airbrush on a spectrogram, that’s the idea basically. The difference here is that besides the brushes that Photosounder has, you can export the image to Photoshop and do some very precise filtering, for example making a sound feature disappear by hand, enhancing parts of a sound, subtracting to sound as I once did by making the difference between a song’s spectrogram and its instrumental version’s spectrogram to isolate the vocals, experiment with contrast, curves, levels, sharpening, various effects (I’m pretty sure you could for example try the glowing edges again and get a different sounding result).

To illustrate the lossless mode, here’s a segment of dialogue from a 1972 social commentary film in the public domain presented with the lossless mode on and again with it off. The lossless mode sounds exactly like the original waveform, while without the lossless mode the audio lacks resolution.

Photosounder Dialogue with Lossless Mode On

Photosounder Dialogue with Lossless Mode Off