The latest articles on DEV Community by Boobalan Arjunan (@boobalan_arjunan). https://dev.to/boobalan_arjunan https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4184493%2F31d4087b-c115-4a5d-bb32-982835587c7d.jpg https://dev.to/boobalan_arjunan en Boobalan Arjunan Wed, 14 Oct 2026 11:09:26 +0000 https://dev.to/boobalan_arjunan/wavewash-clean-repair-and-dedupe-audio-datasets-before-ml-training-4jk4 https://dev.to/boobalan_arjunan/wavewash-clean-repair-and-dedupe-audio-datasets-before-ml-training-4jk4 <p>Training speech models (Whisper, TTS, voice cloning) often breaks down because of dirty audio data: clipping, DC offset, background noise, low SNR, and duplicate samples quietly inflating validation metrics.</p> <p>I built <strong>WaveWash</strong> to automate audio quality control and data prep so developers don't have to write one-off scripts or manually listen to thousands of audio files.</p> <h2> What WaveWash Does </h2> <ul> <li> <strong>Tuned Presets:</strong> Built-in profiles for ASR (Whisper), TTS, Speaker ID, and Music Generation.</li> <li> <strong>Safe Auto-Repairs:</strong> Corrects DC offset, mild clipping, and normalizes loudness (EBU R128).</li> <li> <strong>Automated Segregation:</strong> <ul> <li> <code>clean/</code> : Audio that passed all checks.</li> <li> <code>_error/</code> : Rejected files categorized by reason (<code>low_snr/</code>, <code>clipping/</code>, etc.).</li> <li> <code>_duplicates/</code> : Exact and near-duplicates set aside (nothing is deleted).</li> <li> <code>_logs/</code> : Complete audit logs (CSV, JSON, and Markdown summaries).</li> </ul> </li> <li> <strong>Zero-Config Local Web UI:</strong> Ships directly inside the Python package — no Node.js build required.</li> <li> <strong>Non-Destructive:</strong> Your original files are never modified.</li> </ul> <h2> Quick Start </h2> <p>Install and launch the local web UI in two commands:</p> <div class="highlight js-code-highlight"> <pre class="highlight plaintext"><code> bash pip install wavewash wavewash app It opens your browser at http://127.0.0.1:8420. You can also run it directly from the terminal: wavewash run --input ./raw_audio --output ./clean_audio --purpose asr Links PyPI: https://pypi.org/project/wavewash/ GitHub: https://github.com/BoobalanArjunan/wavewash It's currently in early alpha (v0.1.0a1). Feedback, issues, and feature suggestions are very welcome! </code></pre> </div> python machinelearning opensource ai