FocalOrder: Focal Preference Optimization for Reading Order Detection

Reading order detection is the foundation of document understanding. Most existing methods rely on uniform supervision, implicitly assuming a constant difficulty distribution across layout regions. In this work, we challenge this assumption by revealing a critical flaw: \textbf{Positional Disparity}, a phenomenon where models demonstrate mastery over the deterministic start and end regions but suffer a performance collapse in the complex intermediate sections. This degradation arises because standard training allows the massive volume of easy patterns to drown out the learning signals from difficult layouts. To address this, we propose \textbf{FocalOrder}, a framework driven by \textbf{Focal Preference Optimization (FPO)}. Specifically, FocalOrder employs adaptive difficulty discovery with exponential moving average mechanism to dynamically pinpoint hard-to-learn transitions, while introducing a difficulty-calibrated pairwise ranking objective to enforce global logical consistency. Extensive experiments demonstrate that FocalOrder establishes new state-of-the-art results on OmniDocBench v1.0 and Comp-HRDoc. Our compact model not only outperforms competitive specialized baselines but also significantly surpasses large-scale general VLMs. These results demonstrate that aligning the optimization with intrinsic structural ambiguity of documents is critical for mastering complex document structures.

PARL: Position-Aware Relation Learning Network for Document Layout Analysis

Document layout analysis aims to detect and categorize structural elements (e.g., titles, tables, figures) in scanned or digital documents. Popular methods often rely on high-quality Optical Character Recognition (OCR) to merge visual features with extracted text. This dependency introduces two major drawbacks: propagation of text recognition errors and substantial computational overhead, limiting the robustness and practical applicability of multimodal approaches. In contrast to the prevailing multimodal trend, we argue that effective layout analysis depends not on text-visual fusion, but on a deep understanding of documents' intrinsic visual structure. To this end, we propose PARL (Position-Aware Relation Learning Network), a novel OCR-free, vision-only framework that models layout through positional sensitivity and relational structure. Specifically, we first introduce a Bidirectional Spatial Position-Guided Deformable Attention module to embed explicit positional dependencies among layout elements directly into visual features. Second, we design a Graph Refinement Classifier (GRC) to refine predictions by modeling contextual relationships through a dynamically constructed layout graph. Extensive experiments show PARL achieves state-of-the-art results. It establishes a new benchmark for vision-only methods on DocLayNet and, notably, surpasses even strong multimodal models on M6Doc. Crucially, PARL (65M) is highly efficient, using roughly four times fewer parameters than large multimodal models (256M), demonstrating that sophisticated visual structure modeling can be both more efficient and robust than multimodal fusion.

Bridging the gap: A comparative exploration of Speech-LLM and end-to-end architecture for multilingual conversational ASR

The INTERSPEECH 2025 Challenge on Multilingual Conversational Speech Language Models (MLC-SLM) promotes multilingual conversational ASR with large language models (LLMs). Our previous SHNU-mASR system adopted a competitive parallel-speech-encoder architecture that integrated Whisper and mHuBERT with an LLM. However, it faced two challenges: simple feature concatenation may not fully exploit complementary information, and the performance gap between LLM-based ASR and end-to-end(E2E) encoder-decoder ASR remained unexplored. In this work, we present an enhanced LLM-based ASR framework that combines fine-tuned Whisper and mHuBERT encoders with an LLM to enrich speech representations. We first evaluate E2E Whisper models with LoRA and full fine-tuning on the MLC-SLM ASR task, and then propose cross-attention-based fusion mechanisms for the parallel-speech-encoder. On the official evaluation set of the MLC-SLM Challenge, our system achieves a CER/WER of 10.69%, ranking on par with the top-ranked Track 1 systems, even though it uses only 1,500 hours of baseline training data compared with their large-scale training sets. Nonetheless, we find that our final LLM-based ASR still does not match the performance of a fine-tuned E2E Whisper model, providing valuable empirical guidance for future Speech-LLM design. Our code is publicly available at https://github.com/1535176727/MLC-SLM.

Rhythm-controllable Attention with High Robustness for Long Sentence Speech Synthesis

Regressive Text-to-Speech (TTS) system utilizes attention mechanism to generate alignment between text and acoustic feature sequence. Alignment determines synthesis robustness (e.g, the occurence of skipping, repeating, and collapse) and rhythm via duration control. However, current attention algorithms used in speech synthesis cannot control rhythm using external duration information to generate natural speech while ensuring robustness. In this study, we propose Rhythm-controllable Attention (RC-Attention) based on Tracotron2, which improves robustness and naturalness simultaneously. Proposed attention adopts a trainable scalar learned from four kinds of information to achieve rhythm control, which makes rhythm control more robust and natural, even when synthesized sentences are extremely longer than training corpus. We use word errors counting and AB preference test to measure robustness of proposed method and naturalness of synthesized speech, respectively. Results shows that RC-Attention has the lowest word error rate of nearly 0.6%, compared with 11.8% for baseline system. Moreover, nearly 60% subjects prefer to the speech synthesized with RC-Attention to that with Forward Attention, because the former has more natural rhythm.

M2-CTTS: End-to-End Multi-scale Multi-modal Conversational Text-to-Speech Synthesis

Conversational text-to-speech (TTS) aims to synthesize speech with proper prosody of reply based on the historical conversation. However, it is still a challenge to comprehensively model the conversation, and a majority of conversational TTS systems only focus on extracting global information and omit local prosody features, which contain important fine-grained information like keywords and emphasis. Moreover, it is insufficient to only consider the textual features, and acoustic features also contain various prosody information. Hence, we propose M2-CTTS, an end-to-end multi-scale multi-modal conversational text-to-speech system, aiming to comprehensively utilize historical conversation and enhance prosodic expression. More specifically, we design a textual context module and an acoustic context module with both coarse-grained and fine-grained modeling. Experimental results demonstrate that our model mixed with fine-grained context information and additionally considering acoustic features achieves better prosody performance and naturalness in CMOS tests.

ECAPA-TDNN for Multi-speaker Text-to-speech Synthesis

In recent years, neural network based methods for multi-speaker text-to-speech synthesis (TTS) have made significant progress. However, the current speaker encoder models used in these methods still cannot capture enough speaker information. In this paper, we focus on accurate speaker encoder modeling and propose an end-to-end method that can generate high-quality speech and better similarity for both seen and unseen speakers. The proposed architecture consists of three separately trained components: a speaker encoder based on the state-of-the-art ECAPA-TDNN model which is derived from speaker verification task, a FastSpeech2 based synthesizer, and a HiFi-GAN vocoder. The comparison among different speaker encoder models shows our proposed method can achieve better naturalness and similarity. To efficiently evaluate our synthesized speech, we are the first to adopt deep learning based automatic MOS evaluation methods to assess our results, and these methods show great potential in automatic speech quality assessment.

Selective Pseudo-labeling and Class-wise Discriminative Fusion for Sound Event Detection

In recent years, exploring effective sound separation (SSep) techniques to improve overlapping sound event detection (SED) attracts more and more attention. Creating accurate separation signals to avoid the catastrophic error accumulation during SED model training is very important and challenging. In this study, we first propose a novel selective pseudo-labeling approach, termed SPL, to produce high confidence separated target events from blind sound separation outputs. These target events are then used to fine-tune the original SED model that pre-trained on the sound mixtures in a multi-objective learning style. Then, to further leverage the SSep outputs, a class-wise discriminative fusion is proposed to improve the final SED performances, by combining multiple frame-level event predictions of both sound mixtures and their separated signals. All experiments are performed on the public DCASE 2021 Task 4 dataset, and results show that our approaches significantly outperforms the official baseline, the collar-based F 1, PSDS1 and PSDS2 performances are improved from 44.3%, 37.3% and 54.9% to 46.5%, 44.5% and 75.4%, respectively.

CNN-based Discriminative Training for Domain Compensation in Acoustic Event Detection with Frame-wise Classifier

Domain mismatch is a noteworthy issue in acoustic event detection tasks, as the target domain data is difficult to access in most real applications. In this study, we propose a novel CNN-based discriminative training framework as a domain compensation method to handle this issue. It uses a parallel CNN-based discriminator to learn a pair of high-level intermediate acoustic representations. Together with a binary discriminative loss, the discriminators are forced to maximally exploit the discrimination of heterogeneous acoustic information in each audio clip with target events, which results in a robust paired representations that can well discriminate the target events and background/domain variations separately. Moreover, to better learn the transient characteristics of target events, a frame-wise classifier is designed to perform the final classification. In addition, a two-stage training with the CNN-based discriminator initialization is further proposed to enhance the system training. All experiments are performed on the DCASE 2018 Task3 datasets. Results show that our proposal significantly outperforms the official baseline on cross-domain conditions in AUC by relative $1.8-12.1$% without any performance degradation on in-domain evaluation conditions.

Joint Weakly Supervised AT and AED Using Deep Feature Distillation and Adaptive Focal Loss

A good joint training framework is very helpful to improve the performances of weakly supervised audio tagging (AT) and acoustic event detection (AED) simultaneously. In this study, we propose three methods to improve the best teacher-student framework of DCASE2019 Task 4 for both AT and AED tasks. A frame-level target-events based deep feature distillation is first proposed, it aims to leverage the potential of limited strong-labeled data in weakly supervised framework to learn better intermediate feature maps. Then we propose an adaptive focal loss and two-stage training strategy to enable an effective and more accurate model training, in which the contribution of difficult-to-classify and easy-to-classify acoustic events to the total cost function can be automatically adjusted. Furthermore, an event-specific post processing is designed to improve the prediction of target event time-stamps. Our experiments are performed on the public DCASE2019 Task4 dataset, and results show that our approach achieves competitive performances in both AT (49.8% F1-score) and AED (81.2% F1-score) tasks.