Search

No-Reference video quality assessment has become a trending and challenging hot topic in estimating perceived quality in audiovisual content. In this paper, we present a proposal to considerably reduce the computational cost of video processing without losing accuracy in QoE estimation. Tests have been performed using the Video-MOS SaaS solution, a hybrid NR-VQA solution based on perceptible video distortions and a machine learning approach. After exploring the spatial and temporal redundancy present in a video sequence, the final approach combines video metric feature extraction in both high and low video resolution, together with a specific frame selection based on a uniform temporal sampling and frame type at the video coding level. An extensive validation with more than 144 hours of audiovisual content from six of the most important HD channels of DTT in Spain demonstrates the validity of the approach, ensuring real-time application on the test device, with computational cost savings of 94.96% and an obtained MOS error of 0.1144, in more than 174000 3-second measurements.

Technology innovation for the sixth generation (6G) era should focus on improving quality of life by addressing societal needs, advancing human experience with the fusion of digital and physical worlds, and achieving a sustainable well-being. The 6G networks are expected to bring higher capacity and coverage combined with dependable real-time properties but must additionally be able to control the balance between quality of service (QoS) and quality of experience (QoE). QoS is the ability to offer reliable performance to connect people and things. QoE is a way to measure the quality of a service as perceived by the end-user, enabling a more customer-centric network design approach for end-user services. To provide excellent QoS, accurate estimation of QoE and corrective measures to adapt QoS to maintain acceptable QoE from the end-user perspective will be needed. In this paper, we propose a linear weighted QoE model in terms of mean opinion score (MOS) and the ability to adapt QoS. We further validate our model with a QoE impairment model and analyze the results with respect to the latency and bit rate metrics, which are critical for our target use case of Virtual Reality (VR) gaming. We also propose to use the QoE model to enable adaptive QoS in the 6G radio access network (RAN), including dynamic learning of QoS metrics.

The ICASSP 2023 Acoustic Echo Cancellation Challenge is intended to stimulate research in acoustic echo cancellation (AEC), which is an important area of speech enhancement and is still a top issue in audio communication. This is the fourth AEC challenge and it is enhanced by adding a second track for personalized acoustic echo cancellation, reducing the algorithmic + buffering latency to 20 ms, as well as including a full-band version of AECMOS (Purin et al., 2020). We open source two large datasets to train AEC models under both single talk and double talk scenarios. These datasets consist of recordings from more than 10,000 real audio devices and human speakers in real environments, as well as a synthetic dataset. We open source an online subjective test framework and provide an objective metric for researchers to quickly test their results. The winners of this challenge were selected based on the average mean opinion score (MOS) achieved across all scenarios and the word accuracy (WAcc) rate.

The diffusion model is capable of generating high-quality data through a probabilistic approach. However, it suffers from the drawback of slow generation speed due to its requirement for many time steps. To address this limitation, recent models such as denoising diffusion implicit models (DDIM) focus on sample generation without explicitly modeling the entire probability distribution, while models like denoising diffusion generative adversarial networks (GAN) combine diffusion processes with GANs. In the field of speech synthesis, a recent diffusion speech synthesis model called DiffGAN-TTS, which utilizes the structure of GANs, has been introduced and demonstrates superior performance in both speech quality and generation speed. In this paper, to further enhance the performance of DiffGAN-TTS, we propose a speech synthesis model with two discriminators: a diffusion discriminator to learn the distribution of the reverse process, and a spectrogram discriminator to learn the distribution of the generated data. Objective metrics such as the structural similarity index measure (SSIM), mel-cepstral distortion (MCD), F0 root mean squared error (F0- RMSE), phoneme error rate (PER), word error rate (WER), as well as subjective metrics like mean opinion score (MOS), are used to evaluate the performance of the proposed model. The evaluation results demonstrate that our model matches or exceeds recent state-of-the-art models like FastSpeech 2 and DiffGAN-TTS across various metrics. Our code and audio samples are available on GitHub.

Effectively evaluating the perceptual quality of dehazed images remains an under-explored research issue. In this paper, we propose a no-reference complex-valued convolutional neural network (CV-CNN) model to conduct automatic dehazed image quality evaluation. Specifically, a novel CV-CNN is employed that exploits the advantages of complex-valued representations, achieving better generalization capability on perceptual feature learning than real-valued ones. To learn more discriminative features to analyze the perceptual quality of dehazed images, we design a dual-stream CV-CNN architecture. The dual-stream model comprises a distortion-sensitive stream that operates on the dehazed RGB image, and a haze-aware stream on a novel dark channel difference image. The distortion-sensitive stream accounts for perceptual distortion artifacts, while the haze-aware stream addresses the possible presence of residual haze. Experimental results on three publicly available dehazed image quality assessment (DQA) databases demonstrate the effectiveness and generalization of our proposed CV-CNN DQA model as compared to state-of-the-art no-reference image quality assessment algorithms. [ABSTRACT FROM AUTHOR]