Summary of Multi-modal Retrieval For Large Language Model Based Speech Recognition, by Jari Kolehmainen et al.
Multi-Modal Retrieval For Large Language Model Based Speech Recognition
by Jari Kolehmainen, Aditya Gourav, Prashanth Gurunath Shivakumar, Yile Gu, Ankur Gandhe, Ariya Rastrow, Grant Strimel, Ivan Bulyko
First submitted to arxiv on: 13 Jun 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Sound (cs.SD); Audio and Speech Processing (eess.AS)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes two novel approaches for multi-modal retrieval: kNN-LM and cross-attention techniques. By incorporating external information from multiple modalities, these methods can improve language models’ performance in various machine learning tasks. The authors demonstrate the effectiveness of their approaches by applying them to automatic speech recognition (ASR) tasks, achieving up to 50% improvement in word error rate over a baseline multi-modal language model. Furthermore, they achieve state-of-the-art results on the Spoken-Squad question answering dataset. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps improve language models by adding external information from different sources like text and speech. It proposes two new ways to do this: kNN-LM and cross-attention techniques. The researchers test these methods on speech recognition tasks and show that they can make big improvements, up to 50% better than usual. They also get the best results ever on a dataset called Spoken-Squad. |
Keywords
» Artificial intelligence » Cross attention » Language model » Machine learning » Multi modal » Question answering
