Scholarlink

In this paper, we present a new approach to text generation using a Bidirectional Long Short-Term Memory (LSTM) network. The model is trained on articles from Medium, focusing on predicting the next word in a given text sequence based on an initial input. Before training, the data is carefully processed by tokenizing the text and creating pairs of inputs and outputs for the model. The trained LSTM model delivers highly accurate predictions, showcasing its effectiveness in natural language processing tasks. Additionally, we integrate the text prediction system with Google Text-to-Speech (gTTS) to convert the generated text into spoken language that closely resembles human speech. This combination of text generation and speech synthesis enhances accessibility for visually impaired individuals and creates a more engaging experience in voice-enabled applications. Our system achieves high accuracy, operates efficiently with minimal computational resources, and performs in real-time. These qualities make it a promising solution for assistive technologies and other applications requiring low-resource natural language processing.

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

Real-Time Text Prediction and Speech Synthesis for Accessibility: A Bidirectional LSTM Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Sornavalli Ganesan,
  • Selcia Selvam,
  • Sasmitha Baskaran,
  • Shashwat Shivam

摘要

In this paper, we present a new approach to text generation using a Bidirectional Long Short-Term Memory (LSTM) network. The model is trained on articles from Medium, focusing on predicting the next word in a given text sequence based on an initial input. Before training, the data is carefully processed by tokenizing the text and creating pairs of inputs and outputs for the model. The trained LSTM model delivers highly accurate predictions, showcasing its effectiveness in natural language processing tasks. Additionally, we integrate the text prediction system with Google Text-to-Speech (gTTS) to convert the generated text into spoken language that closely resembles human speech. This combination of text generation and speech synthesis enhances accessibility for visually impaired individuals and creates a more engaging experience in voice-enabled applications. Our system achieves high accuracy, operates efficiently with minimal computational resources, and performs in real-time. These qualities make it a promising solution for assistive technologies and other applications requiring low-resource natural language processing.