# Hidacs Sàrl

> Hidacs Sàrl is a Swiss consulting firm based in Lausanne specializing in audio signal processing, machine learning, and embedded DSP. Founded in 2016 by Patrick Marmaroli (PhD EPFL). Serving clients across Switzerland, France, and the EU. From R&D proof-of-concept to deployed cloud platforms.

Contact: info@hidacs.ch — Lausanne, Switzerland. Engagements start with a free 30–45 min exploratory call, then a short deliverable-based pilot project, with an optional monthly retainer.

Every page exists in three languages behind a language prefix: English `https://hidacs.ch/en/…`, French (default) `/fr/…`, German `/de/…`. A path with no prefix redirects (301) to the French version. The English URLs are listed below.

## Company

- [Home](https://hidacs.ch/en/): R&D verticals, track record, and contact. Four domains — Speech & Voice (clinical speech analysis, anti-spoofing, ASR feasibility), Industrial (acoustic quality control, anomaly detection), Environmental (urban noise monitoring, soundscape classification), Health & Safety (acoustic biomarkers, field event detection).
- [How we work](https://hidacs.ch/en/engagement): The three engagement models — free exploratory call, pilot project, monthly retainer.
- [Case studies](https://hidacs.ch/en/case-studies): Featured case study (clinical voice analysis SaaS — 170+ acoustic ML modules, GDPR-compliant Swiss infrastructure, AI-driven therapy planning) plus five reference projects: acoustic quality control for precision manufacturing (Geneva), acoustic measurement for a watchmaking group (Geneva), smart-city noise monitoring (Geneva), infant feeding audio analysis (FDA Breakthrough Device, New York), acoustic event detection for humanitarian field monitoring (Switzerland).
- [Publications](https://hidacs.ch/en/publications): 13 peer-reviewed papers and patents co-authored by Patrick Marmaroli — sound source localization, indoor acoustic event detection, binaural hearing aids, peripersonal space, clinical speech AI — with DOIs and links.

## Tools (free, public, no signup)

- [All tools](https://hidacs.ch/en/tools): Index of the free audio, DSP and ML web tools below.
- [Digital Audio Filter Designer](https://hidacs.ch/en/tools/audio-filter-design): Interactive FIR filter design for embedded systems. Upload audio, tune parameters, export C headers for Cortex-M/DSP. WASM-accelerated.
- [ML Audio Architecture Assistant](https://hidacs.ch/en/tools/ml-audio-architecture-assistant): AI-powered recommendations for ML audio architectures, with research-based scoring and benchmarks.
- [Swiss Noise Law Assistant](https://hidacs.ch/en/tools/swiss-noise-law-assistant): Compliance guidance on Swiss noise regulations for acoustic projects.
- [SOTA Survey Generator](https://hidacs.ch/en/tools/sota-survey): Generate structured state-of-the-art research surveys with AI. PDF/TXT/JSON export, EN/FR.
- [dB SPL ↔ dBFS Converter](https://hidacs.ch/en/tools/db-converter): Convert between sound pressure level and digital full scale, with microphone sensitivity correction and a reference table for common recording scenarios.
- [GeoKluster](https://hidacs.ch/en/tools/geokluster): Cluster construction site addresses by geographic proximity (k-means), assign teams, export PDF report.
- [SharePoint Remediation Guide](https://hidacs.ch/en/tools/sharepoint-remediation): Interactive step-by-step guide to fix SharePoint governance gaps. 12 issue categories, PowerShell scripts.

## Workshops

- [Interactive Acoustic Workshops](https://hidacs.ch/en/ateliers): Turnkey acoustic experience modules for museums, exhibitions, schools, and events. 11 modules including Chladni Figures, Real-Time Spectrogram, Fourier Wheel, Binaural Listening, ML Classification, Voice Print, Voice Cloning, The Scream. Rental (500–1,500 CHF/week/module) or purchase.

## Blog

Each article has its own URL of the form `https://hidacs.ch/en/blog/<post-id>`.

- [Blog index](https://hidacs.ch/en/blog): All technical articles on signal processing, audio ML, MLOps, and acoustic monitoring.
- [Voice Quality Metrics: Jitter, Shimmer, HNR, and AVQI Explained](https://hidacs.ch/en/blog/voice-quality-metrics) — 2026-06-20 — Clinical voice quality metrics with normal ranges, Praat extraction, and Python code for speech-language pathology pipelines.
- [dB SPL vs dBFS: Understanding Audio Levels for Engineers and Clinicians](https://hidacs.ch/en/blog/db-spl-dbfs) — 2026-06-19 — Conversion formula, calibration workflow, and why the distinction matters for clinical voice analysis and audio ML pipelines.
- [Fine-Tuning Audio ML Models: Plan, Audit, Diagnose](https://hidacs.ch/en/blog/audio-ml-fine-tuning) — 2026-03-01 — Framework for audio ML fine-tuning: plan a new pipeline, audit an existing script, diagnose an underperforming model.
- [MLOps Setup – OVHcloud + DVC + MLflow (Audio Dataset)](https://hidacs.ch/en/blog/mlops-setup) — 2026-02-01 — Reproducible MLOps pipeline for audio datasets: object storage, DVC dataset versioning, MLflow tracking.
- [The Silent Sentinels: When Materials Speak, Quality Control Listens](https://hidacs.ch/en/blog/silent-sentinels) — 2026-01-01 — Four applications of passive acoustic monitoring in manufacturing, from 3D-printed concrete to cryogenic aerospace composites.
- [Wav2Vec2 & XLSR Model Guide](https://hidacs.ch/en/blog/wav2vec2-guide) — 2025-11-01 — Wav2Vec2 variants, use cases, and best practices for speech processing tasks.

## Optional

- [Full content for LLMs](https://hidacs.ch/llms-full.txt)
- [Sitemap (XML)](https://hidacs.ch/sitemap.xml)
- [LinkedIn — Hidacs Sàrl](https://www.linkedin.com/company/hidacs/)
- [LinkedIn — Patrick Marmaroli](https://www.linkedin.com/in/pmarmaroli/)
- [Google Scholar — Patrick Marmaroli](https://scholar.google.com/citations?user=Qa_P6n8AAAAJ&hl=en)
- [ResearchGate — Patrick Marmaroli](https://www.researchgate.net/profile/Patrick-Marmaroli)
- [ORCID — 0000-0001-7359-8670](https://orcid.org/0000-0001-7359-8670)