Summary of A Comprehensive Solution to Connect Speech Encoder and Large Language Model For Asr, by Van Tung Pham et al.
A Comprehensive Solution to Connect Speech Encoder and Large Language Model for ASR
by Van Tung Pham, Yist Lin, Tao Han, Wei Li, Jun Zhang, Lu Lu, Yuxuan Wang
First submitted to arxiv on: 25 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 The paper presents a comprehensive solution to address limitations in connecting speech encoders to large language models (LLMs) for speech recognition. Specifically, it investigates fine-tuning schemes, proposes a matching loss to enhance alignment between modalities, and explores training and inference methods to mitigate high insertion errors. The results demonstrate that partially fine-tuning the encoder and LLM using parameter-efficient methods like LoRA is cost-effective, while the matching loss improves modality alignment and enhances performance. Additionally, the proposed training and inference methods significantly reduce insertion errors. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper solves some big problems in making computers understand what people are saying. Right now, it’s hard to make computers recognize speech when they’re not used to hearing certain types of voices or accents. The researchers came up with three main ideas to fix this: better ways to fine-tune the computer models, a new way to make sure the computer understands both speech and text, and methods to reduce mistakes when the computer is trying to understand what someone said. They tested their ideas on a big database of speeches and found that they really work! This could be super helpful for people who need computers to recognize speech, like those who are deaf or hard of hearing. |
Keywords
» Artificial intelligence » Alignment » Encoder » Fine tuning » Inference » Lora » Parameter efficient
