−18 dBFS, −12 dBFS, −20 dBFS: Reference Levels, dBu and dBV
Where −18 and −20 dBFS come from, why −12 dBFS is a habit, and how to convert dBFS to dBu and dBV for recording levels.
Read articleBlog
Notes & perspectives
Technical notes and practical insights on signal processing, AI and software development.
Articles in English
LUFS vs dBFS: Can You Convert LUFS to dB?
Why there is no LUFS to dB formula, the test signal where there is one, how to measure loudness, and the broadcast and streaming targets.
Read articledB SPL to Pascal: Formula, Table and Microphone Sensitivity
Convert dB SPL to pascals and back: formulas, a reference table, why calibrators use 94 dB, and mV/Pa to dBV/Pa microphone sensitivity.
Read articleHow Construction Joint Ventures Control Costs Across Partner Companies
Why spreadsheets and one partner’s ERP fail a construction consortium, and the ten things a shared cost-control system must do.
Read articleVoice Quality Metrics: Jitter, Shimmer, HNR, and AVQI Explained
Jitter, shimmer, HNR and AVQI: definitions, Praat manual reference values, the AVQI v03.01 equation and Python code with parselmouth.
Read articleFine-Tuning Audio ML Models: Plan, Audit, Diagnose
A structured framework for audio ML fine-tuning: how to plan a new pipeline, audit an existing script, and diagnose an underperforming model.
Read articleMLOps Setup – OVHcloud + DVC + MLflow (Audio Dataset)
A practical guide to a reproducible MLOps pipeline for audio data: OVHcloud Object Storage, DVC for dataset versioning, and MLflow for tracking.
Read articleThe Silent Sentinels: When Materials Speak, Quality Control Listens
Four unexpected uses of passive acoustic monitoring in manufacturing: from 3D-printed concrete to cryogenic aerospace composites, materials speak through sound.
Read articleWav2Vec2 & XLSR Model Guide
A comprehensive guide to Wav2Vec2 model variants, their use cases, and best practices for speech processing tasks.
Read article