Mourning the death of a loved one represents one of the greatest trials of the human experience, and the desire for an enduring connection with them is an intensely personal, unfulfilled aspiration. MemorialBot is an conversational AI that seeks to satisfy this yearning by allowing mourners to interact via natural voice conversation with an artificial reconstruction of a deceased loved one, derived from audio recordings, text documents, and contributions of biographical narratives from the user and loved one. The proposed system architecture comprises an integrated speech-to-speech pipeline for automatic speech recognition (ASR), retrieval-augmented large language model (LLM) inference, neural text-to-speech (TTS) generation, and voice cloning. The performance evaluation included twenty participants who rated voice naturalness at 4.2/5.0, persona consistency at 82% accuracy, and perceived comfort with the system at 4.1/5.0 on a 5-point Likert scale. The paper describes the complete system design, method of implementation, experimental evaluation, and an ethical framework considering consent, data privacy and protection, and emotional safety. MemorialBot offers contributions to the rapidly developing fields of affective computing, digital heritage conservation, and AI support for grief.

In this paper we present a framework and design of MemorialBot-an AI-based system that aims to bridge the gap and lack of engaging, personalized digital memorialization by combining voice cloning, retrieval-augmented LLM inference, and neural TTS within one coherent pipeline that is both usable and interactive. Sentiment-aware response conditioning and governance mechanisms, along with the development of a user-friendly voice-first interface ensures that the system caters to the needs of users while retaining the ethicality that this domain requires. In the larger scope, the framework signifies the capacity for AI technology to facilitate human grief and the progression of affect technology and the digital heritage field.

1. Y. Mori and T. Nakamura, “Digital immortality and AI-based posthumous interaction: a systematic review,” Journal of Human-Computer Studies, vol. 175, p. 103012, 2023. 2.T. Brown et al., “Language models are few-shot learners,” Advances in Neural Information Processing Systems, vol. 33, pp. 1877–1901, 2020. 3.C.J. Sofka, I.N. Cupit, and K.R. Gilbert, Eds., Dying, Death and Grief in an Online Universe. New York: Springer, 2012. 4. G. Park, J. Chung, and S. Lee, “AI chatbot emotional disclosure for mental health counselling,” Current Psychology, vol. 41, no. 8, pp. 5387–5396, 2022. 5. C.H. Reddy, “Real-time voice cloning using generative adversarial networks,” International Journal of Intelligent Systems and Applications in Engineering, vol. 12, no. 2, pp. 245–253, 2024. 6.H. Papneja and N. Yadav, “Self-disclosure to conversational AI: a systematic review,” Personal and Ubiquitous Computing, vol. 29, no. 1, pp. 88–104, 2025. 7. E. Ankomah and R.E. Turkson, “Emotion-aware AI chatbots for mental health support,” World Journal of Public Health, vol. 3, no. 1, pp. 12–21, 2025. 8.S. Kim and H. Kim, “Grief chatbots and emotional dependency: risks and mitigations,” Computers in Human Behavior, vol. 142, p. 107658, 2023. 9. E. Casanova et al., “YourTTS: towards zero-shot multi-speaker TTS and zero-shot voice conversion,” in Proc. ICML, PMLR 162, pp. 2709–2720, 2022. 10.A. Jain and R. Gupta, “Voice-based digital heritage preservation using deep learning,” IEEE Transactions on Affective Computing, vol. 14, no. 3, pp. 1124–1135, 2023.

© 2025 International Journal of Engineering and Techniques (IJET).

Digital Object Identifier (DOI)

IJET assigns a unique DOI to every accepted article via Zenodo for persistent linking and citation. Your article’s DOI: https://doi.org/{{doi}}

Open DOI About Zenodo DOI

Close