arXiv subjects

Haifeng Luo

Publications and source records attributed to Haifeng Luo.

8 recordsLinked to original sources

arXiv2026

Optimising Neural Speech Codecs for 300bps Communication using Reinforcement Learning

In bandwidth-constrained communication such as satellite and underwater channels, speech must often be transmitted at ultra-low bitrates where intelligibility is the primary objective. At such extreme compression levels, codecs trained with acoustic reconstruction losses tend to allocate bits to perceptual detail, leading to substantial degradation in word error rate (WER). This paper proposes ClariCodec, a neural speech codec operating at 300 bits per second (bps) that reformulates quantisation as a stochastic policy, enabling reinforcement learning (RL)-based optimisation of intelligibility. Specifically, the encoder is fine-tuned using WER-driven rewards while the acoustic reconstruction pipeline remains frozen. Even without RL, ClariCodec achieves 4.64% WER on the LibriSpeech test-clean set at 300 bps, already competitive with codecs operating at higher bitrates. Further RL fine-tuning reduces WER to 3.55% on test-clean, corresponding to a 23.5% relative reduction while preserving perceptual quality. In addition, we adapt ClariCodec to a streaming configuration and show that the proposed RL-based optimisation remains effective under streaming constraints, achieving 4.53% WER on test-clean with a theoretical latency of 374 ms.

cs.SD↗

arXiv2026

ClariCodec: Optimising Neural Speech Codes for 200bps Communication using Reinforcement Learning

In bandwidth-constrained communication such as satellite and underwater channels, speech must often be transmitted at ultra-low bitrates where intelligibility is the primary objective. At such extreme compression levels, codecs trained with acoustic reconstruction losses tend to allocate bits to perceptual detail, leading to substantial degradation in word error rate (WER). This paper proposes ClariCodec, a neural speech codec operating at 200 bit per second (bps) that reformulates quantisation as a stochastic policy, enabling reinforcement learning (RL)-based optimisation of intelligibility. Specifically, the encoder is fine-tuned using WER-driven rewards while the acoustic reconstruction pipeline remains frozen. Even without RL, ClariCodec achieves 3.68% WER on the LibriSpeech test-clean set at 200 bps, already competitive with codecs operating at higher bitrates. Further RL fine-tuning reduces WER to 3.20% on test-clean and 8.93% on test-other, corresponding to a 13% relative reduction while preserving perceptual quality.

cs.SD↗

arXiv2025

Audio-Visual Speech Enhancement: Architectural Design and Deployment Strategies

Real-time audio-visual speech enhancement (AVSE) is a key enabler for immersive and interactive multimedia services, yet its performance is tightly constrained by network latency, uplink capacity, and computational delay. This paper presents the design, deployment, and evaluation of a complete cloud-edge-assisted AVSE system operating over a public 5G edge network. The system integrates CNN-based acoustic enhancement and OpenCV-based facial feature extraction with an LSTM fusion network to preserve temporal coherence, and is deployed on a Vodafone-compatible AWS Wavelength edge cloud. Through extensive stress testing, we analyze end-to-end performance under varying network load and adaptive multimedia profiles. Results show that compute placement at the network edge is critical for meeting real-time coherence constraints, and that uplink capacity is often the dominant bottleneck for interactive AVSE services. Only 5G and wired Ethernet consistently satisfied the required communication delay bound for uncompressed audio-video chunks, while aggressive compression reduced payload sizes by up to 80% with negligible perceptual degradation, enabling robust operation under constrained conditions. We further demonstrate a fundamental trade-off between processing latency and enhancement quality, where reduced model complexity lowers delay but degrades reconstruction performance in low-SNR scenarios. Our findings indicate that public 5G edge environments can sustain real-time, interactive AVSE workloads when network and compute resources are carefully orchestrated, although performance margins remain tighter than in dedicated infrastructures. The architectural insights derived from this study provide practical guidelines for the design of delay-sensitive multimedia and perceptual enhancement services on emerging 5G edge-cloud platforms.

cs.SD↗

arXiv2025

Unpacking Graduate Students' Learning Experience with Generative AI Teaching Assistant in A Quantitative Methodology Course

The study was conducted in an Advanced Quantitative Research Methods course involving 20 graduate students. During the course, student inquiries made to the AI were recorded and coded using Bloom's taxonomy and the CLEAR framework. A series of independent sample t-tests and poisson regression analyses were employed to analyse the characteristics of different questions asked by students with different backgrounds. Post course interviews were conducted with 10 students to gain deeper insights into their perceptions. The findings revealed a U-shaped pattern in students' use of the AI assistant, with higher usage at the beginning and towards the end of the course, and a decrease in usage during the middle weeks. Most questions posed to the AI focused on knowledge and comprehension levels, with fewer questions involving deeper cognitive thinking. Students with a weaker mathematical foundation used the AI assistant more frequently, though their inquiries tended to lack explicit and logical structure compared to those with a strong mathematical foundation, who engaged less with the tool. These patterns suggest the need for targeted guidance to optimise the effectiveness of AI tools for students with varying levels of academic proficiency.

cs.HC↗

arXiv2024

Joint Power Allocation and Beamforming for In-band Full-duplex Multi-cell Multi-user Networks

This paper investigates a robust joint power allocation and beamforming scheme for in-band full-duplex multi-cell multi-user (IBFD-MCMU) networks. A mean-squared error (MSE) minimization problem is formulated with constraints on the power budgets and residual self-interference (RSI) power. The problem is not convex, so we decompose it into two sub-problems: interference management beamforming and power allocation, and give closed-form solutions to the sub-problems. Then we propose an iterative algorithm to yield an overall solution. The computational complexity and convergence behavior of the algorithm are analyzed. Our method can enhance the analog self-interference cancellation (ASIC) depth provided by the precoder with less effect on the downlink communication than the existing null-space projection method, inspiring a low-cost but efficient IBFD transceiver design. It can achieve 42.9% of IBFD gain in terms of spectral efficiency with only antenna isolation, while this value increases to 60.9% with further digital self-interference cancellation (DSIC). Numerical results illustrate that our algorithm is robust to hardware impairments and channel uncertainty. With sufficient ASIC depth, our method reduces the computation time by at least 20% than the existing scheme due to its faster convergence speed at the cost of < 12.5% sum rate loss. The benefit is much more significant with single-antenna users that our algorithm saves at least 40% of the computation time at the cost of < 10% sum rate reduction.

cs.IT↗

arXiv2024

On the Secrecy Rate of In-Band Full-duplex Two-way Wiretap Channel

In this paper, we consider a two-way wiretap Multi-Input Multi-Output Multi-antenna Eve (MIMOME) channel, where both nodes (Alice and Bob) transmit and receive in an in-band full-duplex (IBFD) manner. For this system with keyless security, we provide a novel artificial noise (AN) based signal design, where the AN is injected in both signal and null spaces. We present an ergodic secrecy rate approximation to derive the power allocation algorithm. We consider scenarios where AN is known and unknown to legitimate users and include imperfect channel information effects. To maximize secrecy rates subject to the transmit power constraint, a two-step power allocation solution is proposed, where the first step is known at Eve, and the second step helps to improve the secrecy further. We also consider scenarios where partial information is known by Eve and the effects of non-ideal self-interference cancellation. The usefulness and limitations of the resulting power allocation solution are analyzed and verified via simulations. Results show that secrecy rates are less when AN is unknown to receivers or Eve has more information about legitimate users. Since the ergodic approximation only considers Eves distance, the resulting power allocation provides secrecy rates close to the actual ones.

cs.IT↗

arXiv2024

Evaluating Large Language Models in Analysing Classroom Dialogue

This study explores the application of Large Language Models (LLMs), specifically GPT-4, in the analysis of classroom dialogue, a crucial research task for both teaching diagnosis and quality improvement. Recognizing the knowledge-intensive and labor-intensive nature of traditional qualitative methods in educational research, this study investigates the potential of LLM to streamline and enhance the analysis process. The study involves datasets from a middle school, encompassing classroom dialogues across mathematics and Chinese classes. These dialogues were manually coded by educational experts and then analyzed using a customised GPT-4 model. This study focuses on comparing manual annotations with the outputs of GPT-4 to evaluate its efficacy in analyzing educational dialogues. Time efficiency, inter-coder agreement, and inter-coder reliability between human coders and GPT-4 are evaluated. Results indicate substantial time savings with GPT-4, and a high degree of consistency in coding between the model and human coders, with some discrepancies in specific codes. These findings highlight the strong potential of LLM in teaching evaluation and facilitation.

cs.CL↗

arXiv2021

Design and Analysis of Wideband In-Band-Full-Duplex FR2-IAB Networks

This paper develops a 3GPP-inspired design for the in-band-full-duplex (IBFD) integrated access and backhaul (IAB) networks in the frequency range 2 (FR2) band, which can enhance the spectral efficiency (SE) and coverage while reducing the latency. However, the self-interference (SI), which is usually more than 100 dB higher than the signal-of-interest, becomes the major bottleneck in developing these IBFD networks. We design and analyze a subarray-based hybrid beamforming IBFD-IAB system with the RF beamformers obtained via RF codebooks given by a modified Linde-Buzo-Gray (LBG) algorithm. The SI is canceled in three stages, where the first stage of antenna isolation is assumed to be successfully deployed. The second stage consists of the optical domain (OD)-based RF cancellation, where cancelers are connected with the RF chain pairs. The third stage is comprised of the digital cancellation via successive interference cancellation followed by minimum mean-squared error baseband receiver. Multiuser interference in the access link is canceled by zero-forcing at the IAB-node transmitter. Simulations show that under 400 MHz bandwidth, our proposed OD-based RF cancellation can achieve around 25 dB of cancellation with 100 taps. Moreover, the higher the hardware impairment and channel estimation error, the worse digital cancellation ability we can obtain.

cs.IT↗