Audio Transfer Learning with Scikit-learn and Tensorflow

We present a didactic toolkit to rapidly prototype audio classifiers with pre-trained Tensorlow models and Scikit-learn. We use pre-trained Tensorflow models as audio feature extractors, and Scikit-learn classifiers are employed to rapidly prototype competent audio classifiers that can be trained on a CPU.

Check it on Github!

This material was prepared for teaching Tensorflow, Scikit-learn, and deep learning in general. Besides, due to the simplicity of Scikit-learn, this toolkit can be employed to easily build proof-of-concept models with your own data.



Slides: Training neural audio classifiers with few data

During the last summer, I have been a research intern at Telefónica Research (Barcelona). The article “Training neural audio classifiers with few data” is the outcome of this short (but intense!) collaboration with Joan Serrà, where we explored how to train deep learning models with just 1, 2 or 10 audios per class. Check it out on arXiv, and reproduce our results running our code! These slides are the extended version of what I will be presenting next week in ICASSP! See you in Brighton 🙂

Download the slides!

Next step: Dolby Laboratories

After five years at the Music Technology Group, I’m pretty excited to announce that I have accepted a full-time position as a researcher at Dolby Laboratories in Barcelona! I’m very happy to see how professional opportunities around machine learning keep growing in Barcelona. Our beloved city is becoming, more and more, an important AI hub in south Europe — what’s great, because it allows me to keep doing my job while being close to my friends and family!

Continue reading