Large language models (LLMs) have revolutionized natural language processing (NLP) with impressive performance across various text-based tasks. However, the extension of text-dominant LLMs to with speech generation tasks remains underexplored. In this work, we introduce a text-to-speech (TTS) system powered by a fine-tuned Llama model, named TTS-Llama, that achieves state-of-the-art speech synthesis performance. Building on TTS-Llama, we further propose MoLE-Llama, a text-and-speech multimodal LLM developed through purely late-fusion parameter-efficient fine-tuning (PEFT) and a mixture-of-expert architecture. Extensive empirical results demonstrate MoLE-Llama’s competitive performance on both text-only question-answering (QA) and TTS tasks, mitigating catastrophic forgetting issue in either modality. Finally, we further explore MoLE-Llama in text-in-speech-out QA tasks, demonstrating its great potential as a multimodal dialog system capable of speech generation.

Tokenising continuous speech into sequences of discrete tokens and modelling them with language models (LMs) has led to significant success in text-to-speech (TTS) synthesis. Despite these models can generate speech with high quality and naturalness, their synthesised samples can still suffer from artefacts, mispronunciation, word repeating, etc. In this paper, we argue these undesirable properties could partly be caused by the randomness of sampling-based strategies during the autoregressive decoding of LMs. Therefore, we look at maximization-based decoding approaches and propose Temporal Repetition Aware Diverse Beam Search (TRAD-BS) to find the most probable sequences of the generated speech tokens. Experiments with two recent LM-based TTS models demonstrate that our proposed maximisation-based decoding strategy generates speech with fewer mispronunciations and improved speaker consistency1.

We introduce Lumina-Image 2.0, an advanced text-to-image generation framework that achieves significant progress compared to previous work, Lumina-Next. Lumina-Image 2.0 is built upon two key principles: (1) Unification - it adopts a unified architecture (Unified Next-DiT) that treats text and image tokens as a joint sequence, enabling natural cross-modal interactions and allowing seamless task expansion. Besides, since high-quality captioners can provide semantically well-aligned text-image training pairs, we introduce a unified captioning system, Unified Captioner (UniCap), specifically designed for T2I generation tasks. UniCap excels at generating comprehensive and accurate captions, accelerating convergence and enhancing prompt adherence. (2) Efficiency - to improve the efficiency of our proposed model, we develop multi-stage progressive training strategies and introduce inference acceleration techniques without compromising image quality. Extensive evaluations on academic benchmarks and public text-to-image arenas show that Lumina-Image 2.0 delivers strong performances even with only 2.6B parameters, highlighting its scalability and design efficiency. We have released our training details, code, and models at https://github.com/Alpha-VLLM/Lumina-Image-2.0.

In this work, we take the first exploration of the recently popular foundation model, *i.e.,* State Space Model/Mamba, in image quality assessment (IQA), aiming at observing and excavating the perception potential in vision Mamba. A series of works on Mamba has shown its significant potential in various fields, *e.g.,* segmentation and classification. However, the perception capability of Mamba remains under-explored. Consequently, we propose QMamba by revisiting and adapting the Mamba model for three crucial IQA tasks, *i.e.,* task-specific, universal, and transferable IQA, which reveals its clear advantages over existing foundational models, *e.g.,* Swin Transformer, ViT, and CNNs, in terms of perception and computational cost. To improve the transferability of QMamba, we propose the StylePrompt tuning paradigm, where lightweight mean and variance prompts are injected to assist task-adaptive transfer learning of pre-trained QMamba for different downstream IQA tasks. Compared with existing prompt tuning strategies, our StylePrompt enables better perceptual transfer with lower computational cost. Extensive experiments on multiple synthetic, authentic IQA datasets, and cross IQA datasets demonstrate the effectiveness of our proposed QMamba.

Video quality assessment tasks rely heavily on the rich features required for video understanding, such as semantic information, texture, and temporal motion. The existing video foundational model, InternVideo2, has demonstrated strong potential in video understanding tasks due to its large parameter size and large-scale multimodal data pertaining. Building on this, we explored the transferability of InternVideo2 to video quality assessment under compression scenarios. To design a lightweight model suitable for this task, we proposed a distillation method to equip the smaller model with rich compression quality priors. Additionally, we examined the performance of different backbones during the distillation process. The results showed that, compared to other methods, our lightweight model distilled from InternVideo2 achieved excellent performance in compression video quality assessment.

Deep neural networks (DNNs) have shown great potential in no-reference image quality assessment (NR-IQA). However, the annotation of NR-IQA is labor-intensive and time-consuming, which severely limits its application, especially for authentic images. To relieve the dependence on quality annotation, some works have applied unsupervised domain adaptation (UDA) to NR-IQA. However, the above methods ignore the fact that the alignment space used in classification is sub-optimal, since the space is not elaborately designed for perception. To solve this challenge, we propose an effective perception-oriented unsupervised domain adaptation methodStyleAM(StyleAlignment andMixup) for NR-IQA, which transfers sufficient knowledge from label-rich source domain data to label-free target domain images. Specifically, we find a more compact and reliable space i.e., feature style space for perception-oriented UDA based on an interesting observation, that the feature style (i.e., the mean and variance) of the deep layer in DNNs is exactly associated with the quality score in NR-IQA. Therefore, we propose to align the source and target domains in a more perceptual-oriented space i.e., the feature style space, to reduce the intervention from other quality-irrelevant feature factors. Furthermore, to increase the consistency (i.e., ordinal/continuous characteristics) between quality score and its feature style, we also propose a novel feature augmentation strategy Style Mixup, which mixes the feature styles (i.e., the mean and variance) before the last layer of DNNs together with mixing their labels. Extensive experimental results on many cross-domain settings (e.g., synthetic to authentic, and multiple distortions to one distortion) have demonstrated the effectiveness of our proposed StyleAM on NR-IQA.

Short-form UGC video platforms, like Kwai and TikTok, have been an emerging and irreplaceable mainstream media form, thriving on user-friendly engagement, and kaleidoscope creation, etc. However, the advancing content-generation modes, e.g., special effects, and sophisticated processing workflows, e.g., de-artifacts, have introduced significant challenges to recent UGC video quality assessment: (i) the ambiguous contents hinder the identification of quality-determined regions. (ii) the diverse and complicated hybrid distortions are hard to distinguish. To tackle the above challenges and assist in the development of short-form videos, we establish the first large-scale Kwai short Video database for Quality assessment, termed KVQ, which comprises 600 user-uploaded short videos and 3600 processed videos through the diverse practical processing workflows, including preprocessing, transcoding, and enhancement. Among them, the absolute quality score of each video and partial ranking score among indistinguish samples are provided by a team of professional researchers specializing in image processing. Based on this database, we propose the first short-form video quality evaluator, i.e., KSVQE, which enables the quality evaluator to identify the quality-determined semantics with the content understanding of large vision language models (i.e., CLIP) and distinguish the distortions with the distortion understanding module. Experimental results have shown the effectiveness of KSVQE on our KVQ database and popular VQA databases. The project can be found at https://lixinustc.github.io/projects/KVQ/.

We introduce an enhanced spatial perception module, as shown in Fig. 1 , pre-trained on multiple image quality assessment datasets, and a lightweight temporal fusion module to address the no-reference visual quality assessment (NR-VQA) task. This model implements Swin Transformer V2 [1] as a local-level spatial feature extractor and fuses these multi-scale features to enhance the quality-aware information. Furthermore, a temporal transformer is utilized for spatiotemporal feature fusion. To accommodate compressed videos of varying bitrates, we incorporate a coarse-to-fine contrastive strategy, that is, the group contrast loss is used for coarse discrimination of different bitrates, and the rank loss is used at a fine-grained level to enrich the model’s capability to discriminate different quality level.

User Generated Content (UGC) videos are susceptible to complicated and variant degradations and contents, which prevents the existing blind video quality assessment (BVQA) models from good performance since the lack of the adaptability of distortions and contents. To mitigate this, we propose a novel prior-augmented perceptual vision transformer (PriorFormer) for the BVQA of UGC, which boots its adaptability and representation capability for divergent contents and distortions. Concretely, we introduce two powerful priors, i.e., the content and distortion priors, by extracting the content and distortion embeddings from two pre-trained feature extractors. Then we adopt these two powerful embeddings as the adaptive prior tokens, which are transferred to the vision transformer backbone jointly with implicit quality features. Based on the above strategy, the proposed PriorFormer achieves state-of-the-art performance on three public UGC VQA datasets including KoNViD-1K, LIVE-VQC and YouTube-UGC.

Blind Image Quality Assessment (BIQA) is susceptible to poor transferability when the distribution shift occurs, e.g., from synthesis degradation to authentic degradation. To mitigate this, some studies have attempted to design unsupervised domain adaptation (UDA) based schemes for BIQA, which intends to eliminate the domain shift through adversarial-based feature alignment. However, the feature alignment is usually taken at the low-frequency space of features since the global average pooling operation. This ignores the transferable perception knowledge in other frequency components and causes the sub-optimal solution for the UDA of BIQA. To overcome this, from a novel frequency perspective, we propose an effective alignment strategy, i.e., Frequency Alignment (dubbed FreqAlign), to excavate the perception-oriented transferability of BIQA in the frequency space. Concretely, we study what frequency components of features are more proper for perception-oriented alignment. Based on this, we propose to improve the perception-oriented transferability of BIQA by performing feature frequency decomposition and selecting the frequency components that contained the most transferable perception knowledge for alignment. To achieve a stable and effective frequency selection, we further propose the frequency movement with a sliding window to find the optimal frequencies for alignment, which is composed of three strategies, i.e., warm up with pre-training, frequency movement-based selection, and perturbation-based finetuning. Extensive experiments under different domain adaptation settings of BIQA have validated the effectiveness of our proposed method.

For many challenging tasks there is often limited data to train the systems in an end-to-end fashion, which has become increasingly popular for deep-learning. However, these tasks can normally be split into multiple separate modules, with significant quantities of data associated with each module. Spoken language processing applications fit into this scenario, as they usually start with a speech recognition module, followed by multiple task specific modules to achieve the end goal. This work examines how the best use can be made of limited end-to-end training for sequence-to-sequence tasks. The key to improving the use of the data is to more tightly integrate the modules via embeddings, rather than simply propagating words between modules. In this work speech translation is considered as the spoken language application. When significant quantities of in-domain, end-to-end data is available, cascade approaches operate well. When the in-domain data is limited, how-ever, tighter integration between modules enables better use of the data to be made. One of the challenges with tighter integration is how to ensure embedding consistency between the modules. A novel form of embedding-passing between modules is proposed that shows improved performance over both cascade and standard embedding-passing approaches for limited in-domain data.

Spoken language ‘grammatical error correction’ (GEC) is an important mechanism to help learners of a foreign language, here English, improve their spoken grammar. GEC is challeng- ing for non-native spoken language due to interruptions from disfluent speech events such as repetitions and false starts and issues in strictly defining what is acceptable in spoken language. Furthermore there is little labelled data to train models. One way to mitigate the impact of speech events is to use a disflu- ency detection (DD) model. Removing the detected disfluencies converts the speech transcript to be closer to written language, which has significantly more labelled training data. This paper considers two types of approaches to leveraging DD models to boost spoken GEC performance. One is sequential, a separately trained DD model acts as a pre-processing module providing a more structured input to the GEC model. The second approach is to train DD and GEC models in an end-to-end fashion, simul- taneously optimising both modules. Embeddings enable end- to-end models to have a richer information flow. Experimen- tal results show that DD effectively regulates GEC input; end- to-end training works well when fine-tuned on limited labelled in-domain data; and improving DD by incorporating acoustic information helps improve spoken GEC.

Computer assisted language learning (CALL) systems aidlearners to monitor their progress by providing scoring andfeedback on language assessment tasks. Free speaking tests al-low assessment of what a learner has said, as well as how theysaid it. For these tasks, Automatic Speech Recognition (ASR)is required to generate transcriptions of a candidate’s responses,the quality of these transcriptions is crucial to provide reliablefeedback in downstream processes. This paper considers theimpact of ASR performance on Grammatical Error Detection(GED) for free speaking tasks, as an example of providing feed-back on a learner’s use of English. The performance of an ad-vanced deep-learning based GED system, initially trained onwritten corpora, is used to evaluate the influence of ASR errors.One consequence of these errors is that grammatical errors canresult from incorrect transcriptions as well as learner errors, thismay yield confusing feedback. To mitigate the effect of theseerrors, and reduce erroneous feedback, ASR confidence scoresare incorporated into the GED system. By additionally adaptingthe written text GED system to the speech domain, using ASRtranscriptions, significant gains in performance can be achieved.Analysis of the GED performance for different grammatical er-ror types and across grade is also presented.

Assessing pain levels in animals is a crucial, but time-consuming process in maintaining their welfare. Facial expressions in sheep are an efficient and reliable indicator of pain levels. In this paper, we have extended techniques for recognising human facial expressions to encompass facial action units in sheep, which can then facilitate automatic estimation of pain levels. Our multi-level approach starts with detection of sheep faces, localisation of facial landmarks, normalisation and then extraction of facial features. These are described using Histogram of Oriented Gradients, and then classified using Support Vector Machines. Our experiments show an overall accuracy of 67% on sheep Action Units classification. We argue that with more data, our approach on automated pain level assessment can be generalised to other animals.