« What About Bob? | Main | Baby Steps »

Slow Hand Whimsy

Lest you think that I forgot you, I've actually been working.  Well, mostly working.  Sometimes I read technical papers and sometimes I make soup and sometimes I take naps.  I take naps because I get up at 4:30 and I get up at 4:30 because I take naps. 

I've been working on slowing down audio tracks.  As these are primarily audio of someone reading a book, I can take into account both psychoacoustics and human voice modeling.  The former relies upon how our ears (and brain and nervous system, no doubt) work and gets rid of all that pesky aural information you wouldn't hear anyway.  That's why MP3 files can be so much smaller than their CD track counterparts.  Human voice modeling relies upon the knowledge of what your voice box and lips and palate and other soft, mobile parts of your speech apparati can and cannot do.  One thing they do is to create nice waveforms, like the ones below.  The one on the top is Bill Clinton reading from My Life at normal speed.  The one below is the audio file that I have created (algorithmically, I'm not that good a mimic) by the WSOLA algorithm.  That stands for Waveform Similarity OverLay and Add, since you asked.  The basic idea is, if you want to slow down some speech, you walk through the original audio and find a nice little wave, like one of those below.  They're usually quite short, perhaps one ten-thousandth of a second.  If, every once and a while, you duplicate one of these little waves, an amazing thing happens:  you can stretch out the audio without any change in pitch.  It just sounds like Bill is speaking slower. 

Speeding up audio is pretty much the reverse of this process.  You find two wavelets that are next to each other but look similar.  If you get rid of one of them every once and a while . . . voila!  . . . you end up speeding up the speaker.  This actually works pretty well for music, but it takes a lot more processing than I have the horsepower for in the application I'm working on (you have to be pickier about the waveforms and the searches for them take long and so on). 

~~~

I'm doing my own personal version of Slow Food.  I took the potato-leek soup from a couple of days back and added chopped-up kale and red chard (thanks again, Shanna) and a cleavered-off ham hock piece and some more vegetable stock and a few spices.  It's simmering as we speak, so to speak.  I'm going to try to consume it as slowly as Junie eats her dinner in front of House.

That's all for today.  Like I said, I'm pretty busy.  Hope you're all doing well.

TrackBack

TrackBack URL for this entry:
http://www.whimsyspeaks.com/mt-tb.pl/205

Comments

'I take naps because I get up at 4:30 and I get up at 4:30 because I take naps.'

Me too.

Post a comment