Haohan SHI

Mr. Haohan SHI

Institute for Digital Technologies, Loughborough University London, London, City of London, UK
orcid

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
17 February 2024 in Future Internet
Verified
Journal Article
Open access
PDF file
Haohan SHIHaohan SHI
Haohan SHI
Xiyu ShiXiyu Shi
Xiyu Shi
Safak DoganSafak Dogan
Safak Dogan
Audio inpainting plays an important role in addressing incomplete, damaged, or missing audio signals, contributing to improved quality of service and overall user experience in multimedia communications over the Internet and mobile networks. This paper presents an innovative solution for speech inpainting using Long Short-Term Memory (LSTM) networks, i.e., a restoring task where the missing parts of speech signals are recovered from the previous information in the time domain. The lost or corrupted speech signals are also referred to as gaps. We regard the speech inpainting task as a time-series prediction problem in this research work. To address this problem, we designed multi-layer LSTM networks and trained them on different speech datasets. Our study aims to investigate the inpainting performance of the proposed models on different datasets and with varying LSTM layers and explore the effect of multi-layer LSTM networks on the prediction of speech samples in terms of perceived audio quality. The inpainted speech quality is evaluated through the Mean Opinion Score (MOS) and a frequency analysis of the spectrogram. Our proposed multi-layer LSTM models are able to restore up to 1 s of gaps with high perceptual audio quality using the features captured from the time domain only. Specifically, for gap lengths under 500 ms, the MOS can reach up to 3~4, and for gap lengths ranging between 500 ms and 1 s, the MOS can reach up to 2~3. In the time domain, the proposed models can proficiently restore the envelope and trend of lost speech signals. In the frequency domain, the proposed models can restore spectrogram blocks with higher similarity to the original signals at frequencies less than 2.0 kHz and comparatively lower similarity at frequencies in the range of 2.0 kHz~8.0 kHz.

2 August 2025 in arXiv
Verified
Preprint
Open access
PDF file
Haohan SHIHaohan SHI
Haohan SHI
Xiyu ShiXiyu Shi
Xiyu Shi
Safak DoganSafak Dogan
Safak Dogan
THTianjin Huang
Tianjin Huang
YZYunxiao Zhang
Yunxiao Zhang
The rise of highly convincing synthetic speech poses a growing threat to audio communications. Although existing Audio Deepfake Detection (ADD) methods have demonstrated good performance under clean conditions, their effectiveness drops significantly under degradations such as packet losses and speech codec compression in real-world communication environments. In this work, we propose the first unified framework for robust ADD under such degradations, which is designed to effectively accommodate multiple types of Time-Frequency (TF) representations. The core of our framework is a novel Multi-Granularity Adaptive Attention (MGAA) architecture, which employs a set of customizable multi-scale attention heads to capture both global and local receptive fields across varying TF granularities. A novel adaptive fusion mechanism subsequently adjusts and fuses these attention branches based on the saliency of TF regions, allowing the model to dynamically reallocate its focus according to the characteristics of the degradation. This enables the effective localization and amplification of subtle forgery traces. Extensive experiments demonstrate that the proposed framework consistently outperforms state-of-the-art baselines across various real-world communication degradation scenarios, including six speech codecs and five levels of packet losses. In addition, comparative analysis reveals that the MGAA-enhanced features significantly improve separability between real and fake audio classes and sharpen decision boundaries. These results highlight the robustness and practical deployment potential of our framework in real-world communication environments.

18 September 2025 in IEEE Transactions on Audio, Speech and Language Processing
Verified
Journal Article
Closed access
Haohan SHIHaohan SHI
Haohan SHI
Xiyu ShiXiyu Shi
Xiyu Shi
Safak DoganSafak Dogan
Safak Dogan
This paper proposes a Packet Loss Concealment (PLC) method for speech codecs based on linear predictive coding, utilizing attention mechanisms and Long Short-Term Memory networks to reconstruct the Linear Predictive Coefficients. A novel multiscale trend-aware multi-head self-attention network is designed to capture the long-term global correlations and short-term local dependencies of speech signals across different time scales, enabling effective global and local receptive fields during the reconstruction of lost packets. A new multiscale Stack Fusion method is introduced to further enhance reconstruction performance. It assigns higher weights to speech frames closer to the lost packets and lower weights to distant ones, enabling effective integration of global and local features across various time scales. Additionally, a tailored loss function is proposed to guide model training by balancing the numerical precision, structural periodicity, and perceptual fidelity. Objective and subjective evaluations consistently indicate that the proposed method sustains robust performance across varying packet loss rates and speaker variability, underscoring its enhanced generalization. The alignment of improvements across multiple evaluation metrics demonstrates that these advancements are architectural rather than dataset-specific. Notably, the proposed method reveals that integrating codec-internal parameters with multiscale temporal modeling provides intrinsic robustness than post-processing PLC methods. Furthermore, the proposed model requires only 0.16 Giga Multiply-Accumulate Operations per second, underscoring its strong potential for high-quality real-time speech communication applications.