One might call it procrastination. We prefer to call it research. It was time to check out our competitors.
We had to be completely honest with ourselves and admit that we might not have been the first to come up with the amazing idea of transcribing whistling into MIDI format. The only question was: had they implemented it well? Was there no need for our award-winning application anymore?
After spending a significant amount of time listing all of our competitors (= scrolling through the first two pages of Google, then prompting ChatGPT), we eagerly downloaded the first application and whistled the first few lines of a popular song into it. (We used the same song for all applications to make the comparison fair.) The conversion took a few seconds, and then there it was…
Oh well. Let us know if you recognize what this sound sample was meant to be! We figured that maybe the application simply did not support whistling, so we tried again, this time by humming:
After a few tries, we concluded what the reviews of the app had already claimed: the accuracy could use some improvements.
This, of course, couldn’t stop us from trying the next candidate. Same steps: downloading the app, whistling, converting, listening…
…and nothing. We tried a bit of “la-la-la” instead…
…which captured some kind of notes.
Before concluding that we are just completely tone-deaf, we downloaded the next candidate. And then magic happened:
Slightly more jazzy than expected, still not recognizable, but at least we had some notes!
Looking at further apps, the pattern repeated: whistling was mostly not supported, but we had some partial successes with humming or la-la-la. At least we got to familiarize ourselves with different UIs and different functionalities. One of the apps had so many features packed into it that we seriously considered whether the problem was on the user side - but we just couldn’t get it working.
After all the failures and partial successes, we realized: if we manage to get accurate pitch detection, we are already ahead of most of our competitors. But if accuracy is possible, why did so many of them struggle to implement it or catch the issue during QA? Could it be that this is just impossible to do? We will see in the next few weeks!
If you would like to see whether we succeed, follow subscribe click the little bell!!
For those who enjoyed our sound samples, here are a few more tries from different apps to have something to listen to on gloomy days: