CV

Work experience

• Oct 2020 - Sept 2025
  • Huawei; A.I. and audio researcher.
  • Extensive experience with neural networks and large language models applied to multiple audio tasks: acoustics, multimodal text-audio retrieval, music transcription, etc.
  • Led research that produced a series of patented neural network modules for efficient harmonic signal processing, highly applicable to voice and music analysis.
  • Developed low-latency DL prototypes to showcase our developed technologies and demos.
  • Developed and trained models specialized in singing voice transcription.
• March 2019 - Oct 2019
  • Erni; Embedded systems firmware testing
  • Executing tests for different devices
• May 2018 - Feb 2019
  • University of Seville; Embedded systems firmware testing
  • Design and implement tests, and use cases.
  • Design of a testing framework for hardware devices.

Education

• Ph.D, “Deep Learning approaches to singing voice transcription in pop music”, Universitat Pompeu Fabra, 2020-2025
• M.S. in Music and Sound Computing, Universitat Pompeu Fabra, 2019-2020
• B.S. in Telecommunications Engineering, Universidad de Sevilla, 2013-2018

Skills

• Signal processing.
• Python.
• Extrovert team worker.
• Excellent presenter, I’m very good at transmitting ideas.
• Spanish (native speaker); English (fluent); German (intermediate)

Publications

Clasificación de géneros musicales mediante técnicas de aprendizaje automático

Perez, Miguel & Sarmiento Vega, María Auxiliadora & Fondón García, Irene (2018). " Clasificación de géneros musicales mediante técnicas de aprendizaje automático " Bachelor Thesis, Universtiy of Seville.

Harmonic Compatibility for Loops in Electronic Music

Perez, Miguel & Ramires, António & Correya, Albin & Font, Frederic (2020). " Harmonic Compatibility for Loops in Electronic Music " master thesis, Universitat Pompeu Fabra.

A Comparison of Pitch Chroma Extraction Algorithms

Perez, Miguel & Kirchhoff, Holger & Serra, Xavier (2022) "A Comparison of Pitch Chroma Extraction Algorithms " Proceedings of the 19th Sound and Music Computing Conference, Saint-Étienne (France).

TriAD: Capturing harmonics with 3D convolutions

Perez, Miguel & Kirchhoff, Holger & Serra, Xavier (2023). "TriAD: Capturing harmonics with 3D convolutions." Proceedings of the 24th Int. Society for Music Information Retrieval Conference, Milan (Italy).

Improving singing voice transcription generalization with AI-generated accompaniments

Perez, Miguel & Kirchhoff, Holger & Grosche, Peter & Serra, Xavier (2025) "Improving singing voice transcription generalization with AI-generated accompaniments " Proceedings of the 31st conference of Multimedia Modeling, Nara (Japan).

Singing Voice Accompaniment Data Augmentation with Generative Models

Perez, Miguel & Kirchhoff, Holger & Grosche, Peter & Serra, Xavier (2025) "Singing Voice Accompaniment Data Augmentation with Generative Models" 2025 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW), Hyderabad, India, 2025, pp. 1-5.

Refining audio-to-score alignment for singing voice transcription

Perez, Miguel & Kirchhoff, Holger & Grosche, Peter & Serra, Xavier (2025) "Refining audio-to-score alignment for singing voice transcription" Proceedings of the 22nd Sound and Music Computing Conference, Graz (Austria).

Service

• ISMIR Reviewer
• ICASSP Reviewer
• EUSIPCO Reviewer
• INTERSPEECH Reviewer