Skip to content
karamouche/noisekit
Model Serving & MLOps
Generates degraded speech datasets for noise‑robust ASR benchmarking using HuggingFace inputs and audiomentations presets
Python
·
Latest v0.1.5 · 1mo ago
Security brief →
Features
-
Creates atomic degradation scenarios (codec, ambient noise, clipping, reverb)
-
Composes multi‑condition compound presets from atomic ones
-
Scores each output with PESQ, SNR, and NISQA metrics
-
Outputs a JSONL manifest ready for HuggingFace AudioFolder loading
No immediate action
v0.1.5
Mixed
·
Mono audio + transcript‑column
No immediate action
v0.1.3
New feature
·
MuLawCompanding transform
No immediate action
v0.1.2
New feature
·
Code of Conduct + Security Policy
No immediate action
v0.1.1
Breaking risk
·
metadata.jsonl + audio presets
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Search tools, categories, lists, and users
Use ↑↓ to navigate, Enter to open, Esc to close
No results for ""
⌘K to open
↑↓ navigate
⏎ open
