Summary of Listen Again and Choose the Right Answer: a New Paradigm For Automatic Speech Recognition with Large Language Models, by Yuchen Hu et al.
Listen Again and Choose the Right Answer: A New Paradigm for Automatic Speech Recognition with Large Language Models
by Yuchen Hu, Chen Chen, Chengwei Qin, Qiushi Zhu, Eng Siong Chng, Ruizhe Li
First submitted to arxiv on: 16 May 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); 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 research paper proposes a novel approach to generative error correction (GER) for automatic speech recognition (ASR). Recent advances in large language models (LLMs) have shown great effectiveness in enhancing ASR results, but current LLM-based GER methods still suffer from limitations. Specifically, they are unaware of the source speech content and often receive redundant information, which can lead to increased miscorrection. The proposed ClozeGER paradigm addresses these issues by introducing a multimodal LLM (SpeechGPT) that receives source speech as input, reformatting GER as a cloze test with logits calibration to simplify and clarify instructions. This approach achieves a breakthrough over vanilla GER on 9 popular ASR datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about finding ways to improve automatic speech recognition technology. Right now, this tech can get words wrong, so researchers are trying to come up with better ways to correct these mistakes. They’re using really powerful language models that can understand and generate text. These models need more information about the original audio recording to make better corrections. The new approach, called ClozeGER, gives the model access to this source speech information and simplifies the process of correcting errors. This makes a big difference in how well the technology works. |
Keywords
» Artificial intelligence » Logits
