● Independent AI evaluation & engineering Barcelona · EN / ES

Applied AI for music and audio businesses.

Make better decisions about your music AI, catalogue and data.

We help music and audio companies identify model weaknesses, improve catalogue intelligence and validate AI opportunities, using measurable results on their own data.

Independent expertise. Measurable results. Clear technical decisions.

01Fixed scopeagreed in writing before work starts
02Your data stays yoursunder NDA, on a sample you choose
03Evidence, with its limitsevery result reported with its uncertainty
04English and Spanishremote, based in Barcelona
01 — Business challenges

Three problems we help music companies solve

Choose the one closest to yours to see what we evaluate and what it can mean for the business.

01 · AI Model Evaluation

Is your AI performing as well as you think?

Benchmark results do not always reflect real-world performance. A model that performs well on standard datasets may struggle with your actual repertoire, recording conditions or user requirements.

We evaluate models against representative data, identify systematic weaknesses and compare performance across relevant segments. Where appropriate, we combine quantitative metrics with expert musical assessment.

Performance by segmentIllustrative
What we evaluate
  • Precision, recall and F1 by relevant segment
  • Performance across genres and recording conditions
  • Agreement between human annotators
  • Error patterns and domain-specific weaknesses
  • Model cost and latency, when relevant
  • Performance against suitable baselines
Business value

Identify costly weaknesses before deployment and make better-informed technology decisions.

Evaluate your AI
02 — Solutions

From evaluation to implementation

Three ways to work together, depending on where you are. Scope, duration and price are agreed for each project.

  1. 01Assess
  2. 02Build
  3. 03Improve
Stage 01Assess

Validate opportunities before investing.

Understand whether an AI approach is technically viable, what data it requires and whether further investment is justified.

  • AI Opportunity AssessmentA focused assessment of the business problem, available data, technical risks and potential approaches.
  • Feasibility SprintA fixed-scope experiment designed to answer a specific technical question using representative data.

OutcomeA clear recommendation supported by evidence.

Discuss an assessment
Stage 02Build

Develop and test solutions on real data.

Turn a validated technical opportunity into a working prototype that can be evaluated and extended.

  • Applied AI PrototypeA working prototype, built and evaluated on your data.

OutcomeA tested solution and a clearer path towards implementation.

Potential deliverables
  • Functional prototype
  • Reproducible evaluation
  • Technical documentation
  • Defined limitations
  • Handover materials
  • Recommendations for further development
Discuss a prototype
Stage 03Improve

Know how your AI performs over time.

Evaluate existing models, identify weaknesses and support informed decisions about improvements and new versions. How an evaluation works.

  • Continuous AI EvaluationRegular, reproducible checks of models in use or supplied by a vendor.

OutcomeOngoing visibility into model quality and performance.

Potential deliverables
  • Scheduled evaluation
  • Regression checks
  • Performance comparisons
  • Error analysis
  • Technical recommendations
  • Periodic reports
Discuss ongoing evaluation
03 — Our approach

Five steps. Clear decisions.

Every engagement is structured around measurable objectives, documented findings and practical recommendations.

  1. 01

    Discover

    Define the problem and success criteria.

  2. 02

    Measure

    Audit the data and establish a baseline.

  3. 03

    Develop

    Build and test a suitable approach.

  4. 04

    Validate

    Evaluate results, limitations and alternatives.

  5. 05

    Deliver

    Provide findings and recommend the next step.

Every engagement produces documented findings and a clear technical recommendation.

Discuss your project
04 — Consultation

Have a music or audio AI challenge?

Let’s determine whether we can help.

Tell us what you’re working on. We’ll review your enquiry and, if there’s a potential fit, arrange a short introductory call to discuss your goals, available data and possible next steps.

Prefer email? laura.ortega.garre@gmail.com

Optional, but helpful

An approximate range helps us recommend the right engagement. Not sure yet? That’s completely fine.