Shennong: A Python toolbox for audio speech features extraction.

We introduce Shennong, a Python toolbox and command-line utility for audio speech features extraction. It implements a wide range of well-established state-of-the-art algorithms: spectro-temporal filters such as Mel-Frequency Cepstral Filterbank or Predictive Linear Filters, pre-trained neural networks, pitch estimators, speaker normalization methods, and post-processing algorithms. Shennong is an open source, reliable and extensible framework built on top of the popular Kaldi speech processing library. The Python implementation makes it easy to use by non-technical users and integrates with third-party speech modeling and machine learning tools from the Python ecosystem. This paper describes the Shennong software architecture, its core components, and implemented algorithms. Then, three applications illustrate its use. We first present a benchmark of speech features extraction algorithms available in Shennong on a phone discrimination task. We then analyze the performances of a speaker normalization model as a function of the speech duration used for training. We finally compare pitch estimation algorithms on speech under various noise conditions.

Learning spectro-temporal representations of complex sounds with parameterized neural networks.

Deep learning models have become potential candidates for auditory neuroscience research, thanks to their recent successes in a variety of auditory tasks, yet these models often lack interpretability to fully understand the exact computations that have been performed. Here, we proposed a parametrized neural network layer, which computes specific spectro-temporal modulations based on Gabor filters [learnable spectro-temporal filters (STRFs)] and is fully interpretable. We evaluated this layer on speech activity detection, speaker verification, urban sound classification, and zebra finch call type classification. We found that models based on learnable STRFs are on par for all tasks with state-of-the-art and obtain the best performance for speech activity detection. As this layer remains a Gabor filter, it is fully interpretable. Thus, we used quantitative measures to describe distribution of the learned spectro-temporal modulations. Filters adapted to each task and focused mostly on low temporal and spectral modulations. The analyses show that the filters learned on human speech have similar spectro-temporal parameters as the ones measured directly in the human auditory cortex. Finally, we observed that the tasks organized in a meaningful way: the human vocalization tasks closer to each other and bird vocalizations far away from human vocalizations and urban sounds tasks.