Build a local, private, multilingual Wispr Flow-style dictation tool.

If you build with AI, you talk to it all day — and typing is the slow part. The useful target is a Wispr Flow-style dictation tool that does not send your speech to the cloud. It should work on a normal laptop, be private by default, and handle mixed-language speech.

Round 1 starts with Hindi+English because that is where many tools fall apart. Win this by making a fast local engine that writes what was actually said, then put it on GitHub. A free, private, mixed-language dictation tool is useful beyond this round.

1. Get plain English right — the leading submissions already beat the open-source baselines we tested on sample clips.
2. Land the final fast — the finished text lands within about 2 seconds of you stopping. That's the product feel to match, except local and offline.
3. Keep both Hindi and English — write what was actually said (don't translate it to English). This is the real test.
4. Stay on the laptop — no internet while it's scored.
5. Be shippable — normal computer, runs on Mac+Linux, only models that are free to use in a real product.
6. Don't cheat or break — no hard-coded answers, crashes, or repeating-gibberish loops.

You can't win on English alone — the Hindi+English clips are the gate. Just wrapping an off-the-shelf model aces English and fails when both languages appear. The exact rules and reference engine are in the reference-bot write-up.

There is one $500 prize, decided by one combined score: 70 points for the final transcript and 30 points for how quickly that final is ready to paste.

Words and language. Does it write what was actually said, without translating the Hindi away? Build it, test it on your own recordings, and submit — we score it on a hidden set.

Speed of the final. How fast a clean, faithful final lands after you stop talking — the target is ~2 seconds. You don't build a server.Our harness plays real speech into your engine at real time (one simple function — no networking) and measures when a stable final comes back. What's scored is that final — not the live preview while you're still talking; whether you reach it by streaming or batch is your call.

Only the final counts. Early drafts, first-word speed, and rewrites while you are speaking receive no points. A blank or hung final fails that clip.

Everything is scored offline on a single pinned machine — an Apple M1 Pro (8-core CPU + Neural Engine), 32 GB, macOS, accelerator on, network blocked — the same box for every entry and for the RambleFix benchmark. Same hardware, warmed up, identical conditions, so speed is purely about your code. The exact latency targets and weights are in the public streaming contract.

How it's ranked & won: one objective score ranks every entry — and every entry is shown on the board, wherever it lands. RambleFix is the prize bar: beat it on the hidden run and you're highlighted as a qualifier.The $500 goes to the top qualifier. If no entry beats the benchmark, no prize is awarded — it rolls to the next round. We pay only for a real step up.

RambleFix is open. Pull it, run it on your own machine, see exactly where it's weak, and build on top. It uses two models racing — a fast draft and a faithful final — both free-to-ship, no API keys, fully local.

# draft path — whisper.cpp small, loopback only
whisper-server -m ggml-small.bin --host 127.0.0.1 --port 8089 &

# faithful final — a public Hinglish model (Apple GPU)
export RAMBLEFIX_WHISPER_CPP_SERVER="http://127.0.0.1:8089/inference"
export RAMBLEFIX_HINGLISH_FINALIZER="python baseline/finalizer_hinglish.py"

Where it's weak — aim here:

• RambleFix is strong, but not unbeatable. The most direct path is a faster final that still keeps both Hindi and English.
• Hindi-heavy and fast switching clips are the risk. That is where the hidden run decides the prize.
• It drops some terms and numbers. The score caps hard on a flipped term or number — hold proper nouns and digits exactly and you gain points it leaves on the table.

Get the code →

• Read the getting-started guide and the build skill (the high-level architecture to follow), fork the template, and test on the included sample clips (English + Hindi+English) with python preview.py — all offline.
• For the speed score, your engine just needs to return a fast final after the audio stops — one simple function, the shape's in the streaming contract; no servers, we drive the timing and feed the audio in real time.
• Declare your models and their licenses — they must be commercial-friendly, so the winning tool can actually be released for free.
• Email your repo to submit@builderr.ai. We clone it and run it offline on the hidden set on the pinned scoring machine (Apple M1 Pro · macOS · network blocked), and you land on the board.

This challenge is sponsored by Amit— he's backing the $500 bounty and built the RambleFix benchmark. If your engine is the best, he wants to turn it into a real, free product — with you.