Now that Amplenote and GitClear have arrived at the point where they regularly win over new customers based on their intrinsic merit, it leaves us with new and interesting prospects for where to turn our focus. Our challenge: how to enable potential customers recognize that these products could change their lives? Our users tell us that we are creating unique value, but it doesn't do any good if the value we're creating is impossibly difficult to find or understand.
So, how to spread our best ideas in the least time to smart users?
First, because research has proven that the rate at which people can ingest information most rapidly is audio:
Consuming videos offers by far the "widest pipe" through which information can be transferred from one brain to another. The audio from a video can convey 400wpm of information by itself when the user changes video playback speed 2x. But even if the user leaves playback at 1x, they'll still be getting 150wpm of information via audio, compared to 75wpm if they are reading technical material.
The "picture is worth 1,000 words" adage has been repeated for ages for a reason. Many advanced topics defy written explanation, yet are trivial to understand when presented with a picture. We find this especially applicable when it comes to helping users understand our most nuanced+cool features like the Commit Activity Browser.
If video is by far the fastest way to transfer ideas between people, what's the fastest way to transfer ideas to a video? 🕵️
When I first discovered Descript, I called it the most exciting app I'd downloaded in at least a year. As I continued using it over the weeks that followed, I became keenly aware that the tool's heritage is a tool to help create podcasts, and the further you get from that use case, the more the warts start to show. My frustration with Descript peaked on a Monday where I started the day with a finished, recorded script, and just needed to polish some of the video content I'd created. It felt like I was 97% done. But then, for the next 8 hours, I fought through app crashes, unpredictable behavior, and lack of functionality to get the last 3% of polish applied. Mercifully, as my "quick morning project" stretched to 3pm, Descript's cloud storage somehow deleted the main video track I had been using. At that point, I had lost the opportunity to finish my first polished Descript creation, which was sad. But I gained freedom from the nagging sunk cost concerns that were leaving me to keep battling Descript to accomplish seemingly-simple tasks like replacing some snippets of video.
That said, while Descript still has work to do if they hope to become the fully functional audio/video editing tool they promise on their home page, they are undoubtedly an amazing product for the specific purpose of getting a solid, recorded script. As one orates their intended script, Descript automatically transcribes it. The transcribed script can then be edited as a text document, and even when merging together sentences where the intonation of the connected words doesn't match, Descript has built tech that almost always ends up sounding like a naturally spoken product, even when it is a frakensteined mess of cut-and-pasted phrases from sentences.
Their "overdub" feature promises to even allow you to add text to the script that had not been spoken in the initial recording. In having attempted to use this feature about 10x so far, I'd estimate about 2x of those attempts have resulted in a high-enough quality vocal to be used in the final-cut. Thus, I would rate this feature as still very work-in-progress as of November 2021, but a cool concept that's occasionally useful, and will become very useful if they can continue to improve its quality (or I learn to use it better).
My current process for recording a five minute script:
Write around 500-1000 words (3-4 paragraphs), which will prove to be somewhat longer than needed so can be edited down once I learn which parts are easiest to speak aloud
Attempt to record the script in Descript. For the first 30 seconds of the video I try to be visible onscreen, which means extra work positioning the script so its close enough to my webcam that I appear to be making full eye contact 👁️ 👁️
Record script. Ensure that I get at least one good read of each paragraph before moving to the next one. Usually this consists of me reading each paragraph about 3-5x with different intonations until I find one I like, at which point I stop, to make it easier to edit afterwards (usually I can default to assuming the final recorded version of the paragraph was the best version). For a five minute video, this might take me 30 mins.
Attempt to use Descript to edit the recorded script into something production ready. Fail miserably in every case I have attempted so far.
Copy the edited Descript transcript from the first take, paste into Amplenote, and rewrite the original script using the lessons learned about what was difficult to explain or sounded unnatural in my first recording attempt
Re-record with the second script and then re-edit, repeating steps #3 and #4. Usually takes another hour-ish.
At my current level of Youtube "proficiency," it takes about five hours of work to record a five minute script. 😅
After the fateful Monday described above, where Descript eventually deleted my video, it was very clear to me that I was never going to get in the same ballpark as a Shu Omi video using Descript to edit. I started by searching the Mac App Store for good video editing software. The highest rated contender was Wondershare Filmora X, so I tried that out for a couple days. In that time, I grew increasing frustrated by the lagginess of my edits, and it seemed very challenging to import new clips and get them aligned and merged together to my tastes.
After a few more hours of Googling and Youtube searching for the best video editing software, I found this explanation of contenders to be credible, useful, and expertly arranged (split between best "free," "under $200" and "$200+" categories). It convinced me to start a trial of Final Cut Pro, and so far it has proven to be an adequately fast and extremely capable solution for mixing together audio scripts from Descript, video clips (usually made with Quicktime video recorder, since it captures richer color than most apps for reasons I don't understand), and new custom media (titles, graphic overlays, etc).
Final Cut Pro is classic expert software. Everything has a hotkey, and until you start to learn the hotkeys, the names of the 93 panes, and the terminology for the concepts it uses, getting stuff done takes awhile. But over about 15 hours using it thus far, I have yet to experience any major frustration, and I have completed two videos already, with hopes to finish two more in the coming days (follow my efforts on the GitClear and Amplenote Youtube channels).
Hope that this quick beginners take on what has worked for me proves useful to anyone else. If this generates any enthusiasm, I'll probably follow up with additional details about the hardware I've tried thus far, and more details about the Final Cut Pro methods that seem to be most applicable to screencast creation.