During my undergrad studies back in the late 90s and early noughties, I would often participate in laborious transcribing group-efforts for certain content-dense modules. During the many hours spent writing down verbatim reproductions of monotonal lectures from a cassette tape to paper, I acquired first-hand experience of how our memory, be it short or long term, differs (more than we are comfortable to admit) from actual events. This fact would highlight itself as soon as I compared the final transcripts to my correspondent class notes or to my memory recollections of the lecture. As a rule, an amazing number of information given in class was lost, presumably because the human brain cannot maintain a constant level of attention, even if we think we are doing so!
My thesis on the MA in Digital Art and Humanities focuses on Internet Art, arguably the latest chapter in the History of Art. Due to its recentness, much of the material on the topic hasn’t yet been published, no canonical works exist, nor is there a common beaten path for a scholar to follow. The overview of the field is more like an overgrown heather where researchers are still finding the way, the narrative path so to speak. Many of the key artists and critics who have published their thoughts and exhibited their art have done so online, in the form of blogs or videos. Youtube is a great resource for the later and has thus helped my research on Internet Art.
Being influenced by my previous experience of data loss between the spoken and written word, I was inspired to transcribe some of the videos available online on Internet and Contemporary Art which I found the most instructive for my research. Although watching a person speak offers a unique way of gleaning information, in the observation of tones, emphasis, facial expressions and other visual signals all of which aid us, a good chunk of actual hard data contained in the words is invariably lost. An appropriate analogy is trying to capture rain outside by placing a cooking pot in your drive-way. No matter how large your cooking pot is, most of the raindrops will fall outside its perimeter. Now exchange the cooking pot for our biological memory bank and the raindrop for words and you get a clear picture.
My first inclination was thus to find a digital tool which would do the transcribing for me directly from the video to text file, from sound into words. Unfortunately, voice recognition software is still at its early stages and would not be able to transcribe verbatim data from noisy conference rooms with background hubbub and hosting speakers with varying voices and accents.
Searching on the DiRT directory for a tool that would aid me in my task, I came across several promising candidates. One of the first ones I found was Transana, a transcribing program being developed by the University of Wisconsin-Madison since 2001. It seemed like a great tool specially created to aid researchers to transcribe and analyse video and audio data, with cool features such linking the transcripts to specific points in the video. My enthusiasm was curbed when I saw that the price for download was 75 dollars for the standard version.
I next turned to my second candidate in line, the open source program SoundScribbler . Developed by Eric Breck in 1998 at the University of Michigan and intended for use in the Michigan Corpus of Academic Spoken English (MICASE) project, this tight little tool wasn’t as fancy as Transana but offered all the features I really needed to get my research done. Soundscribbler basically plays your media files via installed audio codecs (the most common formats are usable) and allows you to manipulate the speed on reproduction. On top of that, Soundscribbler offers a neat feature called walk which allows you to loop a certain segment of the audio over and over until you get the transcribing of that given segment done. All the commands can be made with short-cuts on the keyboard, so in practice I had my Word document open and would press keystroke commands such as cntr-F5 to play the audio, cntr-F6 to pause it, cntrl-F7 to walk etc… It is a very user friendly program with an intuitive interface and I got the job done quite swiftly: I transcribed the Touching the Art – Episode 2 – Postmodernism, Post-Net & the Art Market, a 6 minute video, in less than 40 minutes.
I found no on-line community, articles or reviews for SoundScribbler. It seems like a pretty unknown and undervalued program (despite being quite sustainable and existing for 17 year) offered as a download from Eric Brecks` personal webpage at Michigan University. Due, however to its incredible efficiency, simplicity and almost instantaneous learning curve, I found that these absences were not detrimental to the usage of the tool. The program is released under the GNU General Public License and Eric Brecks offers the source code, written in C programming language (open source) but asks to be sent a copy of any user alteration. Soundscribbler proves that there are great digital tools available out there that are simply out of the spotlight. This program helps the user create his own data as opposed to creating data from given inputs. It is, in trans-humanism terms, a digital extension to our limited biological memories. Once I create a transcription of a given debate, anyone can read and compare it to the video original, thus verifying the whole process.
Coming back to the video itself, I watched it once carefully before using Soundscribbler and transcribing it. As expected, I was amazed as how certain aspects of the video were invisible on a mere first watch.
My memory had told me that certain guests were more interesting and more participative than others and after transcribing I found it was exactly the opposite! In likewise manner, the Interviewer, Casey Jane Ellison, a charming mixture of the wits of Canadian writer Donna Lypchuck with the edgy looks of Elvira, had spouted out some keywords which, for being new to me, had completely passed by undetected. It is easy to see how data can be lost if we rely on notes or memory. Not only that, while transcribing, we can also tap into the subconscious stream of a debate and observe relations and beliefs that the speakers themselves might be unaware. Here lies one of the key functions and usefulness of a simple online tool like SoundScribbler.
On a later stage, I made a word cloud of the transcribed text using tagul.com and visualized the keywords that were spoken in the debate, thus yet again compensating, with another digital tool, for the limitations of my biological analytical capabilities. But that is another tool for another tale.