ElBruno.S1Mini：本地自动清理语音转文字稿的免费工具

⚡ ElBruno.S1Mini - Clean Up Your Transcripts Automatically

🎯 What Does ElBruno.S1Mini Do?

ElBruno.S1Mini is a free tool that automatically cleans up speech-to-text transcripts. If you use voice recognition software, you know the problem: transcripts are full of "um," "uh," repeated words, and missing punctuation. This app fixes all that for you—right on your own computer. No cloud, no sending your files anywhere. Your data stays private.

Think of it as a smart editor that listens to your transcript and makes it sound professional. It runs entirely offline using a small AI model that downloads itself the first time you use it.

✨ Key Features

Removes Fillers – Deletes "um," "uh," "you know," and other hesitation words
Fixes Stutters – Turns "I I I think" into "I think"
Improves Punctuation – Adds periods, commas, and question marks where they belong
100% Local & Private – Runs on your machine, no internet needed after setup
Works with Common Formats – Handles plain text transcripts easily
One-Click Model Setup – Downloads the AI model automatically from HuggingFace
Built for Windows – Simple, no complicated setup required

📥 Download & Install

Step 1: Get the App

👉 Visit this link to download the application

This page shows you the latest release. Look for the file that says something like ElBruno.S1Mini.zip or ElBruno.S1Mini.exe. Download it to your computer.

🛠️ Step 2: Run the Program

If you downloaded a .zip file: Right-click the file and select "Extract All." Choose a folder (like C:\ElBruno), then open that folder and double-click the application file inside.
If you downloaded a .exe file: Double-click it. Windows might ask for permission—click "Yes."
First Time Setup: When the app opens, it will download the AI model (about 200 MB). This happens once and then you're ready to go.

🚀 Quick Start Guide

🖥️ Your First Transcript Cleanup

Open ElBruno.S1Mini – You'll see a simple window with a text box.
Paste or type your raw transcript into the large box.
Click the "Clean" or "Process" button – Usually a big green button.
Wait a few seconds – The app runs the AI model locally. You'll see a progress bar.
Copy your cleaned text – The result appears in a second box. Click "Copy" to save it to your clipboard.

📝 What Kind of Text Can You Use?

Any transcript from:

Zoom or Teams meetings
Otter.ai or other transcription services
Voice memos turned into text
YouTube auto-captions (copy-pasted)
Any messy text with speech artifacts

Example:

Before:

"Um, so like, I I I think we should, you know, maybe consider um, changing the approach a little bit."

After:

"I think we should consider changing the approach."

💡 Tips for Best Results

Keep it in one language – Works best with English text
Use paragraph breaks – If your transcript has speaker labels, leave them on separate lines
Try long transcripts – The tool handles large batches fine
Run it multiple times – For very messy text, a second pass can polish further

❓ Frequently Asked Questions

Is this really free?

Yes. The software is open-source and completely free to use.

Do I need to buy anything?

No. The AI model is free, and the tool uses free Microsoft ONNX Runtime components.

Will it work on Mac?

The current release is Windows-only. For Mac users, check the GitHub source code—it may be compiled manually.

Does it upload my text anywhere?

No. Everything runs locally on your machine. Once the model is downloaded, you can even disconnect from the internet.

What if the model download fails?

Make sure you have a stable internet connection during the first run. If it still fails, check the GitHub Issues page for help.

How big is the download?

The app itself is under 50 MB, and the model is about 200 MB upon first run.

🧰 System Requirements

Operating System: Windows 10 or Windows 11 (64-bit)
Memory: 4 GB RAM minimum, 8 GB recommended
Storage: At least 1 GB free space (for app + model)
Processor: Any modern Intel or AMD CPU (from 2018 onward)
No GPU required – It runs fine on integrated graphics

🛟 Troubleshooting

The app won't start

Re-download the file—it may have been corrupted
Turn off antivirus temporarily and try again
Make sure you have the latest Windows updates

The model won't download

Check your firewall settings—allow ElBruno to connect
Try a wired connection if on Wi-Fi
Restart the app

The output is still messy

Try pasting shorter sections at once
Make sure your text doesn't include time stamps or weird symbols
Run it through again—it often improves after two passes

🔧 Advanced (For Curious Users)

What's Under the Hood?

ElBruno.S1Mini uses a powerful but small AI model called s1-mini (a compact version of Qwen3). It runs through ONNX Runtime GenAI, which Microsoft maintains. This combination makes the tool fast and efficient.

Can I Modify It?

Absolutely. The source code is available on GitHub. You can change settings, retrain on different text, or integrate it into your own projects. Check the README in the repository for technical details.

Where Can I Report Bugs?

Visit the GitHub Issues page. Describe your problem clearly, include your Windows version, and paste output from the error if any.

How Do I Uninstall?

Simply delete the folder where you extracted the files. There's no registry changes. If you want to remove the downloaded model, it's in a models subfolder.

📚 Related Resources

ONNX Runtime – Microsoft's AI acceleration library
HuggingFace – Where the model comes from (search for "s1-mini")
Qwen3 – The base language model family
Microsoft.Extensions.AI – .NET libraries for AI integration

🙏 Thank You for Using ElBruno.S1Mini!

If this tool saved you time, consider:

Starring the repository on GitHub ⭐
Sharing it with colleagues who transcribe interviews
Reporting any issues you find—every bug report helps improve the app

Privacy First: Your transcripts never leave your computer. No analytics, no tracking, no cloud processing. Period.

📦 Changelog Summary (Latest Version)

v1.2 – Improved punctuation handling, faster model loading
v1.1 – Added batch processing for multiple paragraphs
v1.0 – Initial public release

🔗 Quick Downloads

Main Download Page
Source Code
HuggingFace Model: Automatically downloaded on first run

Made with ❤️ for people who hate editing transcripts manually.

Keywords: ai, asr, csharp, dotnet, dotnet10, dotnet8, huggingface, llm, local-ai, microsoft-extensions-ai, nuget, onnx, onnx-runtime, onnxruntime-genai, qwen3, s1-mini, speech-to-text, text-normalization, transcription, whisper