Transfer Learning From Audio-visual Grounding To Speech Recognition

Abstract

Transfer learning aims to reduce the amount of data required to excel at a new task by re-using the knowledge acquired from learning other related tasks. This paper proposes a novel transfer learning scenario, which distills robust phonetic features from grounding models that are trained to tell whether a pair of image and speech are semantically correlated, without using any textual transcripts. As semantics of speech are largely determined by its lexical content, grounding models learn to preserve phonetic information while disregarding uncorrelated factors, such as speaker and channel. To study the properties of features distilled from different layers, we use them as input separately to train multiple speech recognition models. Empirical results demonstrate that layers closer to input retain more phonetic information, while following layers exhibit greater invariance to domain shift. Moreover, while most previous studies include training data for speech recognition for feature extr

Stats

• citations18
• S2 citations—
• github stars0
• HF likes0
• heat score9.59
• arxiv keyhsu2019transfer

Related papers

• Fine-grained Grounding For Multimodal Speech Recognition (2020)5.84
• Bridging The Gap: Using Deep Acoustic Representations To Learn Grounded Language From Percepts And Raw Speech (2021)0.00
• Multimodal Grounding For Sequence-to-sequence Speech Recognition (2018)8.82
• Knowledge Transfer From Large-scale Pretrained Language Models To End-to-end Speech Recognizers (2022)9.41
• Language Learning Using Speech To Image Retrieval (2019)9.41
• Semantic Speech Retrieval With A Visually Grounded Model Of Untranscribed Speech (2017)10.61
• Towards Visually Grounded Sub-word Speech Unit Discovery (2019)9.03
• Leveraging Pretrained Image-text Models For Improving Audio-visual Learning (2023)0.00