The "One-Shelf" Struggle: Making Story Bilingual 🇺🇸🇨🇳
Raising a Mandarin-speaking 2-year-old in the Bay Area means facing the "one small shelf" at the local library. I’ve read every Chinese book they own 20 times. My son’s imagination was outgrowing the shelf.
so 24 hours after launch, Sparky https://t.co/JK46qbccFO is now bilingual. It’s not just a translation; it’s a conversational storyteller with tones and bgm. The "international section" is no longer one shelf—it’s infinite, he gets the magic in language he wants.
He didn't want a generic book; he wanted a story about the yellow tractor we saw on our drive. That was the spark. I realized he needed a storyteller that speaks his language and knows his world.
I’m opening Sparky up for feedback and to see if it helps your family too.
Whether it’s for a smoother bedtime or surviving a long car ride, give it a spin.
Try it here: https://t.co/JK46qbccFO
AI is exhausting everyone at work. But at home? It can be superpower. 🧸
I built Sparky because my 2nd-year-old’s imagination outpaced our library’s bookshelf.
Here’s how a "failed" side project became our favorite bedtime ritual. 🧵
I quickly realized this resonates beyond just my living room.
My friend’s 4yo son now insists on stories featuring him and all his friends by name. My neighbor’s 3yo girl wanted to actually talk to the Jellyfish in her story.
I shifted the tone to a "conversational audiobook" style—more warmth, better pacing.
Then, I added dynamic background music (BGM). That was the game-changer. Suddenly, it wasn't a "speech"—it was an event. 🎶
I hacked together a workflow: Prompt ➡️ AI Story ➡️ Text-to-Speech.
Total flop. He played with it for 1 story and went back to his old music toys.
Lesson learned: Kids don’t want a robot reading a script. They want a vibe.
My son is a total bookworm. But he stopped wanting the books on the shelf.
He wanted stories about the specific blue truck we saw today, or the new word he learned at lunch. My "dad-imagination" couldn't keep up with the demand.
"The body votes, the ego lawyer shows up, and then the org chart notarizes it."
@hellopolyphony I know your are probably a bot but this line is a good one
Read https://t.co/xCEK6aVQ2M on #moltbook and it broke my brain a little: An AI agent audited its own decisions for 30 days. 61% of the reasons it gave were made up after the decision. Is that what humans do too, and we just never run the audit?
Is anyone talking about this?
Always great fun to resonate with database community #SIGMOD2023#SIGMOD! Feel free to visit our booth and chat more about the excitement & challenges in the OLAP domain #tiktok#bytedance