Learning Speech Representations with Variational Predictive Coding
This work provides a principled interpretation for speech representation learning, potentially advancing methods in the field, though it appears incremental as it builds directly on existing HuBERT-based approaches.
The paper tackled the lack of a theoretical foundation for the HuBERT objective in speech representation learning by showing that variational predictive coding underlies it, leading to simple modifications that improved performance on downstream tasks like phone classification and automatic speech recognition.
Despite being the best known objective for learning speech representations, the HuBERT objective has not been further developed and improved. We argue that it is the lack of an underlying principle that stalls the development, and, in this paper, we show that predictive coding under a variational view is the principle behind the HuBERT objective. Due to its generality, our formulation provides opportunities to improve parameterization and optimization, and we show two simple modifications that bring immediate improvements to the HuBERT objective. In addition, the predictive coding formulation has tight connections to various other objectives, such as APC, CPC, wav2vec, and BEST-RQ. Empirically, the improvement in pre-training brings significant improvements to four downstream tasks: phone classification, f0 tracking, speaker recognition, and automatic speech recognition, highlighting the importance of the predictive coding interpretation.