I’ve been listening to an unusual radio station recently, the just-launched Radio Vox Populi. Built by Blogdex creator Cameron Marlow, the site reads aloud the most recent entries from over a million weblogs, using a combination of Perl scripts, voice synthesis software, and web services. I was happy to help out by supplying the transitional sound effects, randomized from a set of six radio samples that complements the broadcast’s lo-fi style. (Read more about the tech behind the site, and the accompanying art installation.)
In lieu of my usual MP3s, I’ve kept the stream running in the background for the last week or so. I feel like I’m eavesdropping in on the general mood and attitude of the entire blogosphere, lovingly read to me by the computer from Wargames.
For more geeky fun, check out the archive of sound clips that accompanies Dennis Klatt’s 1987 paper on the history of speech synthesis. My personal favorites are the Voder, the first-ever electronic speech synthesis, demonstrated at the 1939 World’s Fair, and the first song in synthesized speech, Bell Labs’ “Bicycle Built for Two” from 1961.
That’s really disturbing.
I like when it hits control characters and just keeps on spouting out numbers and “hash” and “pound”
I’ve always wondered what a condensed blogroll would sound like.
…I only wished it sounded more fluid. I also wish there were a way to read along with what was currently playing so that I could better understand what was being said. I think it’s important to know the source of what is being said. half of the time I couldn’t tell if something sounds more like a commercial or sounds more like an actual blog entry.
What *is* cool is how some speakers seem to have an accent that goes along with the blogger’s place of origin. Time to go put another shrimp on the barbie.
Intelligibility is definitely a problem. Some voices are much clearer than others. Also, there can be large discrepancies in volume between voices. I found myself constantly having to turn the volume up and down. Having said that, what a great idea! This could have a lot of potential.
It’s absolutely amazing and thrilling. After the thrill wears off, yes it’s hard to understand. 🙂
I wonder if you guys could hook up with AT&T Labs text-to-speech which has excellent speech synthesis and a variety of voices and accents.
You know the thing is that first cars weren’t Formula One – this is an amazing start – I feel like ramdac – we are evesdropping on the blogosphere – on the collective utterances of untold numbers of people – deeply cool – reminds of Read It To Me (http://www.alexking.org/index.php?content=software/read_it_to_me/content.php) but different.
It wasn’t just “Bell Labs’s” sample, it was done by Max Mathews, for which the “Max” sound synthesis program is named. Max is still alive and is doing amazing work at Stanford’s CCRMA (www-ccrma.stanford.edu) with radio batons. Also, note that the last thing HAL does in the movie 2001 is sing “Daisy” — this was also homage to Max’s incredible achievement, decades ahead of his time.
It kind of reminds me of days in my room as a kid, playing with SAM on my C64…
A very unusual idea, how well is it doing comercially?
It’s not commercial. It’s an art project.