A newer version of the Gradio SDK is available:
6.1.0
title: Baby Noise Cancellation Demo
emoji: 👶🔇
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 4.44.0
app_file: app.py
pinned: false
license: mit
short_description: AI-powered baby noise removal demo with STT comparison
Baby Noise Cancellation Demo
This Gradio Space demonstrates the effectiveness of AI-powered audio noise removal for cleaning recordings with baby crying in the background.
Purpose
Parents using voice technology often face challenges when children start fussing during dictation, making speech-to-text (STT) transcription difficult or impossible. This demo explores whether deep learning models can effectively remove baby noise while preserving speech quality for accurate STT transcription.
Demo Features
- Side-by-side audio comparison: Listen to the original recording (with baby crying) and the AI-processed version
- STT transcripts: Compare Whisper AI transcripts from both audio versions
- Real-world test case: Authentic recording captured during actual dictation with a 3-month-old baby
Technology
This demonstration uses DeepFilterNet2 for audio noise removal, processing the audio to isolate and preserve speech while suppressing baby crying frequencies.
- DeepFilterNet2: Official repository at Rikorose/DeepFilterNet
- Processing Space: Audio processed using drewThomasson/DeepFilterNet2_no_limit
Citation
@inproceedings{schroeter2022deepfilternet2,
title = {{DeepFilterNet2}: Towards Real-Time Speech Enhancement on Embedded Devices for Full-Band Audio},
author = {Schröter, Hendrik and Escalante-B., Alberto N. and Rosenkranz, Tobias and Maier, Andreas},
booktitle={17th International Workshop on Acoustic Signal Enhancement (IWAENC 2022)},
year = {2022},
}
Results
The processed audio shows significant reduction in background baby noise while maintaining speech intelligibility. Both versions are successfully transcribed by Whisper, demonstrating that the noise removal process preserves the essential speech content.
Use Case
This technology has practical applications for:
- Voice-based productivity tools for parents
- Dictation and note-taking applications
- Voice assistants in noisy environments
- Any speech recognition system that needs to handle background noise
This is a proof-of-concept demonstration showing AI-powered audio cleaning for practical voice technology applications.
About
Audio recorded by: Daniel Rosehill - October 28th, 2025
Created by Daniel Rosehill (GitHub) to explore practical solutions for voice technology in real-world parenting scenarios.