HSGAdviser: AI Speech Assistant for Enabling Sustainable Education Solutions

The speech-based chatbot plays a vital role in advancing equality by providing students with timely, equitable, and cost-efficient academic advices. Moreover, the speech chatbot assistant has enhanced its ability to mimic human interaction by integrating state-of-the-art artificial intelligence models, making interactions faster, more engaging, and requiring less effort compared to text-based chatbots. High school students usually experiencing more stress concerning their future careers compared to other students. This situation creates challenges for both students and academic institutions in providing effective and equitable support, especially for schools that cannot afford to hire academic advisers. To address these challenges, we propose a speech-based chatbot named ‘HSGAdviser’ designed to support high school students equally and efficiently. We integrated this chatbot with the pre-trained models from Automatic Speech Recognition (ASR) such as Google Cloud Speech-to-Text, IBM Watson and DeepSpeech instead of training our own model on an audio corpus, as the latter is more time-consuming and resource-intensive compared to using a pre-trained model. We implemented an encoder-decoder architecture utilizing a Bidirectional Long Short-Term Memory (BiLSTM) for the neural network layer. This model incorporates an embedding layer with a 50-dimensional vector, where each word in the vocabulary list is mapped to this vector. This approach enables capturing similarities in meaning and context between words, thereby reducing memory usage and enhancing training performance, making the model more efficient and sustainable. Accordingly, this chatbot can provide additional contributions to sustainable education, including supporting students with 24/7 support services, offering an affordable academic adviser to schools, and replying to students in real-time.