App · Data & Analytics
Audionomy
Curate, inspect and export audio datasets for speech models.
About
A dataset manager for people training speech models. Load recordings, look at waveforms and mel spectrograms, fix metadata, normalize and chunk audio, then export in formats Hugging Face and PyTorch read directly.
What it does
- Waveform and mel-spectrogram views for spotting clipping, silence and noise
- Batch normalization, chunking and metadata editing
- Exports compatible with Hugging Face datasets and PyTorch loaders
How it's built
Signal processing runs in a Python backend (Librosa); the React front end stays responsive on large batches.