How-2

The 'How-2' dataset is a benchmark used to evaluate the performance of models in generating human-like text summaries from spoken content.

Papers using How-2 (8)

Leveraging Large Text Corpora for End-to-End Speech Summarization2023 · 9 citesAVFormer: Injecting Vision into Frozen Speech Models for Zero-Shot AV-ASR2023 · 8 citesXNOR-FORMER: Learning Accurate Approximations in Long Speech Transformers2022 · 2 citesAttention-based Multi-hypothesis Fusion for Speech Summarization2021 · 1 citesAVATAR: Unconstrained Audiovisual Speech Recognition2022 · 1 citesSpeech Summarization using Restricted Self-Attention2021BASS: Block-wise Adaptation for Speech Summarization2023An End-to-End Speech Summarization Using Large Language Model2024