I’m working on a promotional video, or maybe series of videos, for a local condominium complex and it presented itself as the perfect opportunity to try something I’d heard great things about: Avid’s ScriptSync.
The video project is to promote a few available units in this 55 and older condo, and I decided it would be more interesting and believable if there were no voiceover or narration, if instead the “story” was told by some current residents. So I sat down with five residents and interviewed them on camera, asking them questions about why they like living there, hoping I’d be able to string together their responses into something that made sense.
After the interviews were completed the next important thing to do was to transcribe the conversations. Each interview ended up being 30 minutes long, and I knew it would be just too hard to try to start editing right away, no matter how well I logged everything.
I had never myself dealt with transcription before so I was happy to be pointed to Steve Hullfish’s great transcription article on the Pro Video Coalition and it was very helpful in getting me started thinking about this process.
Because I have (and love) After Effects, I also have Premiere Pro installed, so I tried its speech recognition feature. The results, frankly, were comical.
To do this right a proper transcription was in order. First off, I knew I didn’t have the time (or the inclination) to transcribe the interviews myself, but I was curious what the experience would be like. So I downloaded the excellent InqScribe and used it to transcribe a test interview I had shot. It was a valuable experience, I know that if I were in a pinch and needed to transcribe something myself, I could, and this application worked very well.
Armed with the knowledge that this job needed to be left to the pros, I asked on the Avid-L and FCP-L mailing lists as well as on Twitter for recommendations for transcription companies. I received five referrals, so I decided to hire all of the companies to each tackle one of my interviews. I like to give business to companies that friends recommend, and I encourage you to do the same. Here’s who I worked with and how much each charged to transcribe their approximately 30 minute one-on-one interview:
- Diane McCoy – firstname.lastname@example.org – $60
- Diva Transcription Services – www.divatranscription.com – $42
- TypingPlus – email@example.com – $47.25
- Wordwizards – www.wordwizardsinc.com – $56
- Take1 Transcription – www.take1.tv – $64.50
The costs added up, but considering the alternative of me doing a less good job much more slowly, it was definitely worth it.
If you’ve never done the transcription thing before, here’s how it went for me: I exported a QuickTime movie of each interview then compressed it to a small size suitable for sending over the web. Some of the companies provided me with a login to their own FTP server, others I just uploaded to my own site and provided them with a download link. When they were done they sent back a Microsoft Word document. It couldn’t be easier.
I’ve never actually performed a paper cut before, but this situation called for it. I’m sure there are better approaches, but here’s what I did: I printed each of the transcribed interviews onto colored paper, a different color for each interviewee. Then using scissors I quickly went through and snipped out my questions and useless answers, then placed the remaining paragraphs into a theme-specific pile.
I had these large legal size pieces of paper on the counter, each with a heading written at the top… “Community”, “Security”, “Parking”, “Common area” and so on, each with various bits of paper containing a response that vaguely fit that category of conversation. If you stood back and squinted your eyes real tight it vaugely resembled the interface of an NLE– bins filled with clips.
Then I went through each category and refined my cuttings, paring them down to just the bit I was likely to use, then I arranged the responses on the 8.5×14 paper so the ideas kinda flowed, then I taped them down. The result was a rainbow colored script!
So the first part of my editorial process was decidedly low tech, but the next was on the other end of the spectrum. I converted the transcript Word docs into text files, adjusted their formatting and then the real fun began.
In Media Composer (I’m using 4.02) you choose New Script… from the File menu. You’re prompted to select a text file, then a new window appears containing the transcribed text. Select all of the text in this window, or the part that corresponds with the clip you’re about to link with, then drag the master clip or subclip whose audio matches the script onto the the script.
Next comes the real magic sauce. Choose ScriptSync… from the Script menu and the Avid magically listens to your audio media and associates the words on the script to the words said onscreen. It is a fast process too, indexing one of my 30 minute interviews took just 50 seconds.
Now you’re able to search the script window for words or phrases, then once found you double click on the script and the clip opens in the source window with the position indicator parked right where the person is saying the words you clicked on. It is incredible.
So I just went through my taped together script searching the script for a key word in that section and I was very quickly able to create a rough sequence based on my paper cut. It really is amazing.
So that’s that. I’m now in editing mode, cleaning up the bites and covering with b-roll and figuring out what I need to go back and shoot. I couldn’t be this far along this quickly without this amazing feature of Media Composer, check it out for yourself!
Some of these interviewees just aren’t phrasing things as I need. ScriptSync is magic here too, because for example I’m able to search for “s.” to find a sentence that ends with a proper idea-ending “sss” sound.
Original bite as spoken by subject:
I searched for the characters “s.” and quickly found other places in the interview where apparently my interviewee ended a sentence using a word ending in “s”. A quick edit and I have something that sounds much better.
Imagine doing this without ScriptSync, well I know you don’t have to imagine, you’ve done it. So have I. This way is definitely a nice alternative.