Hi guys, thank you for your comments. Bloody hell!
Firstly, I completely understand the initial adverse reaction to anything containing AI – I also tend to disregard most content the moment I sniff AI, and assume it’s just low-quality slop created in an afternoon with the goal of bringing in income.
In light of the discussion which has opened up here, I would like to shed some light on my background, the process of creating the deck, and the rationale for the choices I made along the way, and hopefully convince you that this does not belong in the slop bin.
What initially made me want to create this deck was a general disappointment in the quality and availability of quality German frequency decks. Every one I tried was riddled with mistakes. I say this as a C2 speaker of German with a degree in linguistics, 15 years’ experience learning German and 10 years’ experience as a DE–EN translator. I would consider myself, for all intents and purposes, a native speaker.
I decided I wanted to create a quality deck to plug this gap, and the result is the 1,500 word deck shared here, which took me 6 months and several hundred hours to finish.
I first started by cross-referencing around 15 different frequency lists and then painstakingly adding and subtracting words based on contemporary usage and daily utility. This selection process alone took over 100 hours. I then worked with native speakers to write example sentences for each of the 1,500 words.
Initially, I wanted to get a native speaker to record the sentences in a recording studio. Although I do usually get mistaken for a native speaker, I wanted someone with a standard accent who had acquired the language from early childhood. All of the quotes for a professional voiceover artist were around $1,000, which I was unable to justify as my first side project deck. (It’s perhaps worth mentioning that my second deck, which is slightly more manageable in its scope, does contain native human audio). If I do manage to cover the costs and prove there is demand, then professionally-recorded audio will be absolutely non-negotiable going forward.
I also considered buying audio equipment and setting up my own home studio, but I couldn’t find anybody in my social circle with an appropriate voice who was able or willing to deliver the necessary audio.
After going back and forth and considering my options, I settled on an ElevenLabs voice clone based on a real-life, professional voiceover artist. This is not comparable to the old TTS we were used to from 90% of Anki decks just a few years ago. I showed it to a number of native speakers and none of them could tell that the voice was not “real” and were all impressed by the quality. The almost undetectable difference between this and true native audio is not of any significance in the context of a beginner learning basic vocabulary. If I had not disclosed this, I very much doubt anybody would have questioned it.
I don’t see a problem with using AI sparingly when justified, providing an expert is in control of the final product, which I was. As of 2026, I think that resisting targeted and justified use of AI is going to be a fairly futile exercise going forward.
In terms of the 200 version, I genuinely do believe this is a quality resource for beginners in its own right. I don’t see anything wrong with offering it for free as a way to attract customers to a premium version. This is a side passion project for me, and it barely covers the power costs of using my computer, let alone the time I spent on it, nor even the ElevenLabs tokens I used. I’m certainly not “raking it in at the expense of learners”, as is the implication here.
Anyway, I’m an honest creator with a passion for languages, and I have nothing to hide. If anyone has read this far and wants free access to the full version so they can see that it’s actually a quality product which they would be happy to use themselves, let me know.
If this is against the policy and spirit of this forum, then feel free to delete the deck and continue with your day.
In the spirit of this thread, I’ve written this completely “manually”, though I’m probably going to strongly regret not giving it a onceover with ChatGPT once the pedantic comments start rolling in.