Why I Built This — And Why It's Not Enough Yet
First of all, there is a free tier. You should never be taxed for your hearing.
I'm hard of hearing, a professional software engineer, and an avid gamer. For years, I watched every live captioning solution fail gamers like me - too slow, too inaccurate, too clunky to actually use while playing. So I built something better.
But the real reason I started was my son. A hearing kid getting closer to gaming age. I made myself a promise: we will play online games together, over voice chat, and I won't miss a word.
I spent the last 7 months building CaptionsRush. I reached out to game studios. I tested every speech-to-text model I could find. I dug deep into why real-time captioning keeps failing gamers — and how to fix it. As far as I know, I'm the only hard-of-hearing developer working on a solution like this.
The conclusion I reached: this problem is solvable. The technology exists. What's missing is the will — and the community.
The Two Problems Standing in Our Way
Cost. The STT models that actually work cost money. After extensive testing, Google's model comes out on top — the same engine powering Google Meet. It's fast, accurate, and supports 125 languages. It costs $1 per hour. I offer it through CaptionsRush at zero profit. You pay for what you use. I take nothing.
Games are ignoring our needs. For STT to work well, it needs clean audio. Games don't give us that — explosions, music, ambient sound all bleed into voice chat. Most games also mix all players into a single audio channel, making speaker identification impossible. The STT engine gets a wall of noise and has to fight through it.
Both of these are solvable. The answer is community.
How Community Changes Everything
On cost: When enough users commit to a single platform, I can negotiate volume discounts directly with Google. The math is real:
- 500 users → cost drops from $1.00 to $0.60/hour
- 1,000 users → $0.48/hour
- 2,000 users → $0.24/hour
The bigger the community, the cheaper it gets for everyone. There's also a government angle: subsidies flow toward centralized, verifiable solutions. A fragmented community doesn't give policymakers a pipeline to deliver help through. A unified one does.
On game studios: I've spoken with more than 10 studios. Every single one gave me the same answer: "It's a noble cause — but you don't justify the investment yet. Come back when you have the numbers."
That answer is wrong. And I intend to prove it.
A large, active community is leverage. When a studio knows that meeting our accessibility standards means CaptionsRush recommends their game — driving a wave of HoH players, better Steam rankings, real exposure - the conversation changes. The developers I've spoken to aren't hostile. They're waiting for proof that we exist in numbers.
The first concrete thing I want to push for: a dedicated accessibility option that automatically lowers SFX, music, and ambient audio when someone speaks on voice chat, while keeping the voice channel at full volume. That single change would dramatically improve STT accuracy across every game that implements it.
Longer term, I'm building toward a game engine SDK — similar to how audio middleware like Wwise works — that would let studios plug CaptionsRush directly into specific audio pipelines: isolate voice chat, tag sound events like [footsteps], [explosion], [rain], and deliver the cleanest possible audio for transcription, with full speaker identification. I've done game development. I know this is buildable. But studios will only adopt an SDK when the community justifies it.
What I'm Asking For
I'm not asking for money.
I'm asking for your presence. Install CaptionsRush. Join the Discord. Be part of the number I can show game studios and cloud providers when I walk into the room.
Many people can use AI tools to replicate software. Nobody can use AI to build a real community. That part only we can do -- together.
A note on free alternatives: Windows 11 Live Captions exists, and yes - with borderless mode workarounds, you can force it over a game. But it supports a limited set of languages, it wasn't designed for gaming, and the UI doesn't belong in the middle of a raid. I'm not a native English speaker. My language isn't on Microsoft's list. This isn't a workaround problem. It's a representation problem.