Show HN: Talk to Me Human – my game about social persuasion (talktomehuman.com)
I recently graduated from a good PhD program studying NLP. Unlike any sane person who would go become a professor or make a gazillion dollars in industry, I decided to try bootstrapping my own software business. This is my first product.
The inspiration was from my research on computers understanding social norms. When ChatGPT came out, I was amazed how well it could understand social etiquette. I thought it'd be fun to make a game where you have to talk your way out of sticky situations - like you miss your friend's birthday party, or your boss catches you trying to leave work at 2pm.
I made a prototype in a couple days, and it was super fun to play with. I thought I'd spend a "couple months" making a game for others to play online. Now, only 10 months and 923.3 hours of work later, it's playable in early access.
In the game, you talk out loud (ASR), and the NPCs (LLM + TTS) talk back at you. It is fun to play with a friend! And because it's just talking, non-gamers do great, often better than gamers.
I really want to have a free demo, but no time yet to implement. For now, it's purchase only ($4.99). If anyone decides to try it, I'd really love to get more feedback. It was an enormous learning experience, especially targeting the web - so many partially supported web APIs and browser inconsistencies! Still feels like 2008 in some ways.
Also happy to answer questions of course. Thanks, and enjoy the weekend!
54 comments
[ 4.8 ms ] story [ 110 ms ] threadRecommended for the novelty and very well produced content
This reminds me of "Death by AI"[1], although that one uses text-based input instead of speech.
[1] https://news.ycombinator.com/item?id=38318889
Also, I didn’t realize from the website that it actually uses voice? That should probably be highlighted, unless I just missed it.
One other thought, your title says it’s a game about social persuasion. In the examples on the site it seemed like many of the challenges just involved coming up with lies. If there are no consequences to just making up lies it seems to kind of flatten the strategy. Though I’m just speculating. Thanks for sharing!
Specifically, re: flattening the strategy, spot on and I’m excited to introduce more long term repercussions for lying. There are some long-range dialog repercussions now, but not many. Having to craft a consistent narrative across the whole game (“Overboard” comes to mind if you’ve played that) would be wildly interesting.
Loud and clear re: demo and highlighting voice!
Everything else (graphics, voices, dialogue, interface, story, transcription) worked great and I don't think need to be improved. But the main mechanics of the game feels like a thin layer of LLM on top of a rigid skeleton of scripted scenarios with fixed outcomes. I'd love to see the language model on a longer leash.
Thanks for sharing and huge kudos for creating a playable game that was legitimately enjoyable! I look forward to the next parts of the story.
Writing speech recognition instead of ASR would be clearer to anyone without a similar background.
Your audience is HN so my guess is that you assume we all know the industry/academic terms here. I’m unsure how correct that assumption is - I’m just suggesting take care when using your acronyms in other places.
https://en.wikipedia.org/wiki/Façade_(video_game)
https://youtu.be/nFDsHXSAzs8?si=iN_x-yR5FC9gCuIf
My NLP background left me on a hilariously level playing field with everyone else to basically become a prompt engineer for a while ;)
If anyone is interested in working together, reply. I started working with TTS, ASR, and LLM API's because this piqued my curiosity after attending a few meetups and going in "cold" without much social interaction before the meeting. I imagine a proper app could give people superpowers.
A playable demo will max engagement but you won't see more paying conversions from it. Maybe less.
I don't think I remember a single person, out of hundreds, who didn't subscribe within an hour of trying the software out. I don't think anyone waited for the end of their 30-day trial. I did have a trial, though, not just a video.
But how does it impact your average user rating?