Scholarlink

The popularity of online learning has led to a need for dependable student support systems, which has spurred the development of cutting-edge solutions like chatbots. In this study, a novel approach to creating a “Doubt Solver” chatbot tailored to certain online course materials is presented. We utilize the Whisper library’s automated speech recognition (ASR) features to precisely transcribe instructional videos from various platforms, such as Coursera and YouTube. After that, metadata is added to these transcripts, and they are fed into the potent Language Model (LLM) framework LangChain. LangChain’s few-shot learning algorithms enable the chatbot to provide contextually relevant responses, providing learners with individualized support in real time. By giving students accurate and fast replies, this research shows the effectiveness and feasibility of integrating ASR and few-shot learning to enhance the educational experience. This research highlights the effectiveness of LangChain with a precision of 85% in the ConversationalRetrievalQA chain.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Doubt Solver Chatbot Using Speech-to-Text in Whisper and Few-Shot Learning in LangChain

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Dharin Dave,
  • Yash Agarwal,
  • Harshraj Chouhan,
  • Sakshi Jambhulkar,
  • Santwana Gudadhe,
  • Anuradha Thakare

摘要

The popularity of online learning has led to a need for dependable student support systems, which has spurred the development of cutting-edge solutions like chatbots. In this study, a novel approach to creating a “Doubt Solver” chatbot tailored to certain online course materials is presented. We utilize the Whisper library’s automated speech recognition (ASR) features to precisely transcribe instructional videos from various platforms, such as Coursera and YouTube. After that, metadata is added to these transcripts, and they are fed into the potent Language Model (LLM) framework LangChain. LangChain’s few-shot learning algorithms enable the chatbot to provide contextually relevant responses, providing learners with individualized support in real time. By giving students accurate and fast replies, this research shows the effectiveness and feasibility of integrating ASR and few-shot learning to enhance the educational experience. This research highlights the effectiveness of LangChain with a precision of 85% in the ConversationalRetrievalQA chain.