To address the low accuracy and poor adaptation of generic speech recognition models to unique pronunciations and vocabularies in Chongqing dialect scenarios, this paper constructs a multi-scenario and multi-speaker Chongqing dialect speech dataset. Based on the FunASR framework, we fine-tune the lightweight pretrained model SenseVoiceSmall. Four groups of controlled experiments are designed: baseline, SpecAugment augmentation only, domain hotword enhancement only, and their combination. Results show that the joint optimization strategy achieves the best performance: the model’s Average Correctness (Avg Corr) increases from 81.67% to 84.48%, and Average Character Error Rate (Avg CER) decreases from 25.31% to 20.09%. The recognition of colloquial expressions, unique vocabularies, and typical pronunciations of Chongqing dialect is significantly improved. Meanwhile, the model achieves a Real-Time Factor (RTF) of 0.005 with an average inference latency of 0.045 seconds per utterance, demonstrating high efficiency for lightweight deployment. This work provides a lightweight solution for Chongqing dialect speech recognition and a valuable reference for low-resource dialect research.
Pengcheng Zhou, Yanhua Liang, Dehong He et al.· International Conference on...· 0 citations
With the widespread deployment of face recognition systems, high-fidelity spoofing attacks such as photo replays, video replays, and 3D masks pose significant security threats. Current face detection approaches (e.g., CNN-based methods) often fail to adequately capture the distinct properties of different modalities during feature fusion, resulting in persistent security risks within face detection systems. This paper proposes a multi-dimensional feature fusion-based face anti-spoofing detection method based on the YOLOv8 architecture. Unlike traditional methods that rely solely on static RGB input, the paper integrates dynamic optical flow features (capturing micro-movements) with static texture analysis. By utilizing a cross-scene dataset containing diverse attack samples ,the method achieves robust detection through the fusion of spatial and temporal cues. Additionally, data augmentation strategies are employed to help the model better localize and identify targets across different scales. The proposed method exhibits strong practical value for intelligent systems in complex scenarios.
Yanhua Liang, Pengcheng Zhou, Hongmei Qin et al.· International Conference on...· 0 citations