The Zero Resource Speech Challenge 2021: Spoken language modelling
This addresses the problem of developing speech models without labeled data for researchers in speech processing, though it is incremental as it builds on prior zero-resource challenges.
The paper introduced the Zero Resource Speech Challenge 2021, which tasked participants with learning a language model directly from audio without text or labels, using the Libri-light dataset, and reported results from eight submitted systems evaluated across acoustic, lexical, syntactic, and semantic metrics.
We present the Zero Resource Speech Challenge 2021, which asks participants to learn a language model directly from audio, without any text or labels. The challenge is based on the Libri-light dataset, which provides up to 60k hours of audio from English audio books without any associated text. We provide a pipeline baseline system consisting on an encoder based on contrastive predictive coding (CPC), a quantizer ($k$-means) and a standard language model (BERT or LSTM). The metrics evaluate the learned representations at the acoustic (ABX discrimination), lexical (spot-the-word), syntactic (acceptability judgment) and semantic levels (similarity judgment). We present an overview of the eight submitted systems from four groups and discuss the main results.