You are browsing as a guest. Sign up (or log in) to start making projects!

CleanTone

  • 3 Devlogs
  • 3 Total hours

I am developing a web application called “CleanTone” for improving the audio recordings of instruments/voices captured in regular room settings. The user uploads the audio file, and the application cleans it by decreasing noise levels and making the sound balanced. Currently, I am working on my MVP, which includes audio file upload, simple improvements, and comparison between the original and improved audio files. A few bugs are left to be fixed but it will happen.

Open comments for this post

1h 22m 18s logged

I built CleanTone as a browser-based audio enhancement tool, and in this session I focused on polishing the details: I swapped in my own logo image, added a meta description and favicon so the site looks legitimate when shared or searched, set the license to MIT, and cleaned out dependencies in package.json that were leftover from an earlier version I never finished trimming. I also added a small file-counter feature to the footer myself, but in typing it out by hand I used the wrong quote style and accidentally broke the whole script — every single feature on the page stopped working because of one syntax error. I caught it, found the exact line, fixed it, and verified the whole app worked end-to-end again before pushing. The new logo is attached below plus I hosted it in Vercel so other people can see it. Claude helped me signficantly too in fixing and making various changes and so did Github Copilot in fixing mistakes that Claude made and I also helped in coding significantly.

0

Loading discussion…

0
3
Ship #1 Changes requested

CleanTone is a browser-based audio enhancement app that cleans up instrument recordings in real-time—you upload an MP3 or WAV, click enhance, and it removes background noise, boosts clarity, cuts rumble, and normalizes volume, all without uploading anything to a server. Everything runs locally in your browser using Web Audio API and a custom FFT-based spectral denoiser. The biggest challenge was making the enhancement actually audible—the initial implementation was technically correct but barely noticeable, so I rewrote it to learn from the quietest 12% of frames, which suddenly made the 15-20dB improvement immediately obvious. I'm proud that the audio actually works (that SNR improvement is real, not marketing fluff), your audio never leaves your device (entire processing happens in seconds), and the attention to detail throughout—visual feedback for everything, helpful error messages that tell you exactly what went wrong and how to fix it, and full keyboard support with Ctrl+E to enhance, Ctrl+R to reset, Ctrl+S to download. To test it: clone from https://github.com/manayparikh1/CleanTone, run npm install && npm start, open http://localhost:3001, upload any MP3 or WAV file (phone recordings or live instruments with room noise work best), try different presets (Balanced, Voice, Strings, Wind, Max Clean), A/B compare the waveforms and audio, adjust sliders to fine-tune, click any stat to copy it, and listen for the hiss to vanish while the instrument gets clearer and louder. The app is completely private, open-source, and free—no accounts, no tracking, nothing uploaded anywhere. And AI such as Claude and Github Copilot aided me a lot in debugging and coding etc.

  • 3 devlogs
  • 3h
Try project → See source code →
Open comments for this post

17m 55s logged

Claude and I teamed up to make everything look more good for the user so that it’s user friendly and helpful in the same time.

0

Loading discussion…

0
3
Open comments for this post

1h 34m 19s logged

I am working on the connections to the browser because I want to run this locally instead of purchasing an api. So with Claude we are figuring out on how change acoustics just enough to make it sound more enhanced and better. I completed the dashboard and the features page with the play buttons where individuals could also compare their two audios.

0

Loading discussion…

0
4

Followers

Loading…