Earlier this week I attended the Project Bamboo workshop at Princeton and thought I’d share a discovery I made while trying to complete the pre-workshop reading assignments attendees received:
1. Please read the proposal in its entirety. The proposal can be found at:
http://projectbamboo.org/files/docs/bamboo_proposal.pdf
2. Please read the Identifying Scholarly Practices handout. This handout can be found at:
http://projectbamboo.org/files/3/Identifying_Scholarly_Practices.pdf
All through the week leading up to the workshop I figured I’d surely get around to reading those documents but I never did. The night before I was to leave, I realized it just wasn’t going to happen. Then I had a thought: why not run the text of these documents through my Mac’s “Text to Speech” service, capture the output and later listen to it as a podcast during the dead time of my 3+ hour drive up to New Jersey?
I launched WireTap Studio (to capture the sound) then opened the proposal PDF in Preview, highlighted the text of the document, selected “Services -> Speech -> Start Speaking Text” under Preview’s application menu and hit the record button. I stopped after 30 seconds and imported the mp3 into iTunes. Sounded terrible—lots of ambient noise and sort of muddy sound quality. Oh yeah, I was just picking up the tinny sound of the Macbook’s speakers with the low-quality built-in mic, no wonder it sounded so bad.
Next tried to use WireTap Studio to intercept the audio stream (could also do this with Audio Hijack Pro) and found that I couldn’t seem to interrupt (and grab) the Speech services audio. It wasn’t associated with the application and no matter what I selected as input, it didn’t get the speech audio. I assume it can be done but I wasn’t having any success. Time to Google…
Doh! Turns out there’s a unix command baked right in to OSX (since 10.3) that not only does exactly what I was trying to do, it does is much faster than the real-time capture I was experimenting with. Meet “say”
say [-v voice] [-o out.aiff] [-f file]
So, I opened the PDF in Preview, used Command-A to select all the text, pasted it into a text file using BBEdit, chopped out the parts I didn’t care about then saved it to the desktop. Then in a terminal window, issued this command:
say -v Alex -o ~/desktop/bamboo.aiff -f ~/desktop/bamboo.txt
Alex is the “new and improved” voice in OS X 10.5 (Leopard). He has much better inflection and sounds much more human and much less Cylon. If you really get into this (or need a voice that deals with a language other than US English), you can purchase additional voices from Cepstral (http://www.cepstral.com). The voices are roughly $30 each.
In a little less than 3 minutes wall time, ‘say’ produced the bamboo.aiff file that was easily imported into iTunes (2 hours, 5 minutes of audio). Here’s a representative sample of how Alex sounded with the material:
[audio:bamboo-sample.mp3]…information technologists to collectively tackle the question: How can we enhance arts and humanities research through the development of shared technology services? This proposal represents an 18-month planning and community design program, the Bamboo Planning Project, where through a series of conversations and workshops, we will map out the scholarly practices and common technology challenges across and among disciplines, and discover where a coordinated, cross-disciplinary development effort can best foster academic innovation. Input into the Bamboo process…