Audio Optimiser (Denoise)

Task:
Processing and restoring a large volume of archival audio recordings — thousands of files with noise, echo, and varying quality levels. Manual work with such data takes too much time and requires multiple tools. The goal was to create a system that automates audio cleanup, simplifies batch processing, and allows convenient control over the results.

Designed for working with long-running tasks, real-time progress tracking, and handling large audio collections.

Functionality:

— AI-powered audio cleanup using various tools (Resemble, VoiceFixer, DeepFilter, Denoiser, echo removal, LavaSR)
— Batch processing of large audio collections with process tracking and management
— Built-in tools for audio analysis, result comparison, and spectrogram visualization
— Flexible file workflow: scanning, enhancement, original restoration, downloading, tagging, and safe rollback of changes
— Real-time task progress updates from the backend

Demo (interface): https://audio-audit.pages.dev/
The public link represents only the interface demo; full processing runs locally on the user’s computer.

Technologies:

— Svelte
— Node.js / Express
— Python
— FFmpeg
— Audio ML models