OpenAI engineer posted "One day," and the lightweight CPU-compatible TTS project became a hit.

SnapshotBot

2026-03-29 11:25:01

Abstract generation in progress

A Two-Word Tweet That Sparked Speculation

What Happened

Jason Liu (@jxnlco) replied to a KittenTTS-related account on social media with just two words: “One day.”

Who is Liu? An OpenAI employee, he has written the Instructor structured LLM output library and is quite well-known in the developer community.

In short: This isn’t big news. But for those following lightweight open-source TTS, it’s an interesting little signal.

What is KittenTTS

A lightweight text-to-speech project. The parameter count ranges from 15 million to 80 million, with the smallest model packaged at around 25MB. The selling point of version 0.8 is that it doesn’t require a GPU; it can run on a regular computer CPU.

The code is here: github.com/KittenML/KittenTTS
Liu’s background: Previously worked at Stitch Fix and Meta, the Instructor library has over 6 million monthly downloads, and 11,000 stars on GitHub.

Metric	Data
Parameter Count	15M–80M
Minimum Model Size	About 25MB
Hardware Requirements	Just a CPU

What Does “One day” Really Mean

Honestly, no one knows. The context of the original post is incomplete. But we can guess a few possibilities:

Simply expressing hope: “One day it will be better.”
Implying that OpenAI is internally paying attention to similar directions.
Just a casual response that came to mind, with no other meaning.

Liu has previously posted about coding agents and using AI tools to assist workflows (2026-03-27). His public persona has always revolved around the theme of “lowering barriers.”

Why Some People Care About This

Practical Value: TTS that can run offline on a laptop, without needing to pay for an API or be online, is genuinely appealing for resource-limited developers.
Route Rivalry: Big companies pouring compute power into large models is one path, while lightweight open-source is another; there are people pursuing both.
Signal Value: The post itself doesn’t have high interaction, but what OpenAI engineers are interested in can indicate some underlying issues.

How to View This Matter

Importance: Moderate. What can you analyze from two words? More of a data point for “industry interest distribution.”

Conclusion: An early signal, for reference only. Most relevant to developers and open-source contributors.

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.