Software Engineer at Capital One UK
A look back on the history of my dictation app, Transcrybr. It has been nearly two and a half years since it was launched in September 2020. Today I’m launching V2.0 of the app after adding a sync to iCloud feature (more on this later). This post contains some stories of key new features from the beginning to the V2.0 launch, and some stats at the end.
It’s a dictation or transcription app that generates text from audio. You can review the generated transcript with the text highlighted as the audio is played back. And of course both the audio and transcript can be exported.
In V2.0, the recordings are synced to iCloud which means they will show up in the app on another Apple device you signed in. Find out more at its landing page or download here.
The app had some birth defects. First, I tried to find a problem for a solution. This is something many business/startup gurus advise against. I think the better way is to find a solution for a problem, preferably a problem you have as well so you are the user of the product too.
Anyway, I found this cool dictation solution, Apple’s Speech framework, which takes in audio stream and returns transcription in almost real time. I considered many use cases it could be used for. I contacted a speech therapist to see whether she would find it useful for their patients. She gave it to one of her patients, but I didn’t hear anything back. Eventually, I settled on podcasters. A demographic which needs to transcribe audio to text on a weekly or monthly basis, depending on how often they release new episodes. I talked about this in my launch blog post as well. There was some positive feedback and I was optimistic. But after launch, it wasn’t exactly what I had imagined in the shower. Sales didn’t take off, and the churn rate was high (still high 😅). I quickly lost my motivation to continue after having actively worked on it for about four months. And I went to work on other projects (which also didn’t really work out well).
A few months later, around mid-2021, after failing at a few other projects, Transcrybr still generated around £40 each month and became my most profitable side project, even though I hadn’t been working on it.
So, rather than completely abandon it, I decided to check in on it once every few months, making sure everything still works, tidying things up here and there. And whenever I receive feature requests or bug reports from users (I have a section in the app for users to email me), I work on it.
A short blog post for a 1-year review: https://www.tanin.me/posts/2021/nov/transcrybr-after-1-year-ish/.
The majority of the emails I receive are users asking to cancel the subscription.
And I usually reply with a link to Apple unscribscribing instructions.
There have been a few feature request emails and bug report emails too. Here are the two notable feature requests from users that I ended up adding to the app.
It started from a simple request.
My app previously just chose whichever the system thought was the default mic. And usually, if a pair of Airpods was paired, the Bluetooth headphone would be the default mic. But in this case, it was a dedicated Bluetooth mic, and the app didn’t select it by default. I had thought about the feature to select the mic , but decided to skip it in the first iteration. So it was finally time to work on this feature.
I eventually came up with a solution to show a bottom sheet when a Bluetooth mic is detected.
Later I asked what he was trying to use the app for, and he said it was for his yoga class. He taught senior citizens, some with severe hearing loss. So he needed to use the app to transcribe what he said. Although, he would be standing far apart from the phone with the app running, so he needed to use a Bluetooth mic rather than the phone mic.
We discussed a few options for his use case and I suggested ignoring Transcrybr in favour of Zoom as well 😂.
Eventually, the app was released as I anticipated (something I won’t promise to the user again as Apple approval time can be longer). The user tried it out, but a new issue arose.
The individual was so kind as to explain the situation and note that the new issue wasn’t really an issue with the app. Even though I didn’t solve his problem, the app got slightly better in the process so it’s still a win.
The feature was released in V1.34
This was quite an obvious feature I had been putting off, and again, a user asking for it was a good trigger to get started. There weren’t many email exchanges as it was quite straightforward from the user’s perspective.
The development was rather complex, though, and I ran into some tricky technical challenges that took me a few days to resolve.
Due to its technical challenges, this will be @transcrybr biggest update since its first release 2-ish years ago. I'm working on saving recordings on iCloud and syncing them across devices!☁️ pic.twitter.com/XilBM6WXTp— Tanin (@LandTanin) January 13, 2023
This tweet was posted on 13th January 2023, but it still took me nearly a week (on and off) to polish everything.
Releasing this new feature was a little scary as well, as I only tested it on a few devices that I have (basically my iPhone Xs and iPad mini 5). I initially made the app move any existing local files to iCloud, but later I changed it to just copy existing local files to iCloud to avoid the risk of removing/losing existing files.
There was still some fear that the download feature might not work well, so I put it on a 7-day phased release
And also tracked and monitored the error events related to the iCloud sync feature.
The only error found was basically me testing the error screen, so far so good! (The analytic website tracks sessions anonymously, hence why it assigns a random name to my session -Carlee Konop…)
And with that in mind, I’ll let the phased release finish and therefore complete Transcrybr V2.0 release 🎉
02/2/2023 Edit: post-release reply from the user
Case closed 😌
I shared download and revenue stats in my last blog post. These are two of the most asked questions so if you have skipped the history section, I hope at least you find this section useful!
Last time I shared, it was 2.86K downloads after 1 year. So 5.25K after 2 years and a bit seems like a steady growth. However, churn is high, and the monthly average paying users is only 13 users.
Note: The app is free to download with 15 minutes of free transcription. After that, it’s £3.99/month or £15.49/year (price varies by region).
$1.43K (~£1158) after 2 years and 4 months, for a side project is not bad at all. This means I make ~£40/month or £480/year on average. At least it pays for the annual Apple developer fee of £75. So even if revenue from here doesn’t grow any further, I can release more apps without paying the annual Apple developer fee!
(Or pay for hosting of other projects 💸)
As I was composing the tweet pic.twitter.com/eiIk7nnfXu— Tanin (@LandTanin) January 9, 2023