First introduced in a research paper in late October 2019, our Polyphonic Sound Detection Score (PSDS) is an industry-standard evaluation framework and metric for polyphonic sound recognition systems.

We’ve now made the PSDS source code available on GitHub, enabling researchers to apply the score to their sound recognition models.

Dr Sacha Krstulović, Director of AA Labs (Audio Analytic’s research group), said: “PSDS solves some fundamental shortcomings of previous evaluation approaches, therefore we believe that the wider sound recognition community will benefit from accessing the source code and using PSDS for their own work.”

The PSDS GitHub repository comprises a python package which contains a library that calculates the PSDS of polyphonic sound event detection systems.

Dr Krstulović continues: “Our approach to evaluating the performance of polyphonic sound event detection systems revisits the definition of system errors, makes the evaluation more robust and expands the evaluation to include the factors which matter to user experience, for example the cross-triggers or the stability across classes.

“It means that with PSDS, the identification of the best performing system becomes grounded in user experience rather than abstract or impractical statistical definitions.”

Read more about the Polyphonic Sound Detection Score and download the research paper ‘A Framework for the Robust Evaluation of Sound Event Detection’, which has been submitted to ICASSP 2020.

You can find the full technical paper on the Polyphonic Sound Detection Score, along with links to GitHub and the Jupyter Notebook, here.

We are using our own and third party cookies which track your use of our website to help give you the best experience. By continuing, we’ll assume that you are happy to receive all cookies on our website.

You can check what cookies we use and how to manage them here and you read our privacy policy here.

Accept and close