Task-Agnostic Structured Pruning of Speech Representation Models
This work addresses the industrial applicability of speech models by reducing their size and speed, though it is incremental as it builds on existing pruning techniques.
The paper tackles the problem of large memory and computational requirements of self-supervised speech models by proposing a structured pruning method with fine-grained attention head pruning and L0 regularization, achieving comparable performance to dense models with 72% fewer parameters and 2x faster inference speed on the SUPERB benchmark.
Self-supervised pre-trained models such as Wav2vec2, Hubert, and WavLM have been shown to significantly improve many speech tasks. However, their large memory and strong computational requirements hinder their industrial applicability. Structured pruning is a hardware-friendly model compression technique but usually results in a larger loss of accuracy. In this paper, we propose a fine-grained attention head pruning method to compensate for the performance degradation. In addition, we also introduce the straight through estimator into the L0 regularization to further accelerate the pruned model. Experiments on the SUPERB benchmark show that our model can achieve comparable performance to the dense model in multiple tasks and outperforms the Wav2vec 2.0 base model on average, with 72% fewer parameters and 2 times faster inference speed.