Assessing the perceptual quality of synthetic speech is crucial for guiding the development and refinement of speech generation models. However, it has traditionally relied on human subjective ratings such as the Mean Opinion Score (MOS), which depend on manual annotations and often suffer from inconsistent rating standards and poor reproducibility. To address these limitations, we introduce MOS-RMBench, a unified benchmark that reformulates diverse MOS datasets into a preference-comparison setting, enabling rigorous evaluation across different datasets. Building on MOS-RMBench, we systematically construct and evaluate three paradigms for reward modeling: scalar reward models, semi-scalar reward models, and generative reward models (GRMs). Our experiments reveal three key findings: (1) scalar models achieve the strongest overall performance, consistently exceeding 74% accuracy; (2) most models perform considerably worse on synthetic speech than on human speech; and (3) all models struggle on pairs with very small MOS differences. To improve performance on these challenging pairs, we propose a MOS-aware GRM that incorporates an MOS-difference-based reward function, enabling the model to adaptively scale rewards according to the difficulty of each sample pair. Experimental results show that the MOS-aware GRM significantly improves fine-grained quality discrimination and narrows the gap with scalar models on the most challenging cases. We hope this work will establish both a benchmark and a methodological framework to foster more rigorous and scalable research in automatic speech quality assessment. Assessing the perceptual quality of synthetic speech is crucial for guiding the development and refinement of speech generation models (V alentini-Botinhao & Y amagishi, 2018).

We thank all reviewers for their input on our paper. WaveFlow is a hybrid between autoregressive and parallel flow-based models. Add definition of p (x), q( y) after equation 1. (R3) Done.

This paper introduces a dual-layer framework for network operator-side quality of experience (QoE) assessment that integrates both objective network modeling and subjective user perception extracted from live-streaming platforms. On the objective side, we develop a machine learning model trained on mean opinion scores (MOS) computed via the ITU-T P.1203 reference implementation, allowing accurate prediction of user-perceived video quality using only network parameters such as packet loss, delay, jitter, and throughput without reliance on video content or client-side instrumentation. On the subjective side, we present a semantic filtering and scoring pipeline that processes user comments from live streams to extract performance-related feedback. A large language model is used to assign scalar MOS scores to filtered comments in a deterministic and reproducible manner. To support scalable and interpretable analysis, we construct a labeled dataset of 47,894 live-stream comments, of which about 34,000 are identified as QoE-relevant through multi-layer semantic filtering. Each comment is enriched with simulated Internet Service Provider attribution and temporally aligned using synthetic timestamps in 5-min intervals. The resulting dataset enables operator-level aggregation and time-series analysis of user-perceived quality. A delta MOS metric is proposed to measure each Internet service provider's deviation from platform-wide sentiment, allowing detection of localized degradations even in the absence of direct network telemetry. A controlled outage simulation confirms the framework's effectiveness in identifying service disruptions through comment-based trends alone. The system provides each operator with its own subjective MOS and the global platform average per interval, enabling real-time interpretation of performance deviations and comparison with objective network-based QoE estimates.

Additional Feedback: Comments: - Section 2: Flow-based models are not necessarily large. The new SOTA WaveFlow is a small-footprint flow-based model for raw audio. The authors may reference WaveFlow and clarify the inaccurate claim in related work section. I usually don't take such FDSD measures seriously, as it couldn't provide meaningful comparisons across different models in general, which is also observed by the authors. It would very nice to see an ablation study with MOS scores by varying three design choices: 1) w/ or w/o repulsive term, 2) single or multi-scale spectrogram loss, 3) w/ or w/o GAN loss. It will single out and emphasize the benefit of repulsive term under different circumstances.

Quality: This paper suffers from a few critical issues. Clarity: The experiment setting ups can be described with more details. Sec 3.2 and 3.4 is missing important information such as the datasets used for conducting the experiments. Significance: Although the quality of the proposed model remains unclear because of the previously mentioned critical issues, it's a significant work because it's the first GAN-based model for spectrogram-to-waveform conversion which seems to be working at some degree. It's significantly over-claimed: 1) claiming state-of-the-art for spectrogram-to-waveform conversion (line 6) with MOS 3.09 is surprising; many previous works are at a much higher level (e.g.

The generation of natural and high-quality speech from text is a challenging problem in the field of natural language processing. In addition to speech generation, speech editing is also a crucial task, which requires the seamless and unnoticeable integration of edited speech into synthesized speech. We propose a novel approach to speech editing by leveraging a pre-trained text-to-speech (TTS) model, such as FastSpeech 2, and incorporating a double attention block network on top of it to automatically merge the synthesized mel-spectrogram with the mel-spectrogram of the edited text. We refer to this model as AttentionStitch, as it harnesses attention to stitch audio samples together. We evaluate the proposed AttentionStitch model against state-of-the-art baselines on both single and multi-speaker datasets, namely LJSpeech and VCTK. We demonstrate its superior performance through an objective and a subjective evaluation test involving 15 human participants. AttentionStitch is capable of producing high-quality speech, even for words not seen during training, while operating automatically without the need for human intervention. Moreover, AttentionStitch is fast during both training and inference and is able to generate human-sounding edited speech.

This paper addresses the gap in We are particularly inspired by approaches in biology where efficient audio quality prediction, especially in low-resource zero-shot prediction is possible using a model's uncertainty settings where extensive MOS data from large-scale listening estimates, where uncertainties act as proxies for downstream tests may be unavailable. We demonstrate that uncertainty tasks [4]. Our main hypotheses are that, measures derived from out-of-the-box pretrained selfsupervised learning (SSL) models, such as wav2vec, correlate 1. uncertainty estimates can be derived from the outputs with MOS scores. These findings are based on data from the of SSL models such as wav2vec, and that, 2022 and 2023 VoiceMOS challenges. We explore the extent 2. these uncertainties can be used as proxies to MOS of this correlation across different models and language scores as high model uncertainty around the contents contexts, revealing insights into how inherent uncertainties in of an audio sequence must correspond to low audio SSL models can serve as effective proxies for audio quality quality.

We describe an end-to-end speech synthesis system that uses generative adversarial training. We train our Vocoder for raw phoneme-to-audio conversion, using explicit phonetic, pitch and duration modeling. We experiment with several pre-trained models for contextualized and decontextualized word embeddings and we introduce a new method for highly expressive character voice matching, based on discreet style tokens.

Speech quality assessment has been a critical component in many voice communication related applications such as telephony and online conferencing. Traditional intrusive speech quality assessment requires the clean reference of the degraded utterance to provide an accurate quality measurement. This requirement limits the usability of these methods in real-world scenarios. On the other hand, non-intrusive subjective measurement is the ``golden standard" in evaluating speech quality as human listeners can intrinsically evaluate the quality of any degraded speech with ease. In this paper, we propose a novel end-to-end model structure called Convolutional Context-Aware Transformer (CCAT) network to predict the mean opinion score (MOS) of human raters. We evaluate our model on three MOS-annotated datasets spanning multiple languages and distortion types and submit our results to the ConferencingSpeech 2022 Challenge. Our experiments show that CCAT provides promising MOS predictions compared to current state-of-art non-intrusive speech assessment models with average Pearson correlation coefficient (PCC) increasing from 0.530 to 0.697 and average RMSE decreasing from 0.768 to 0.570 compared to the baseline model on the challenge evaluation test set.

While it might be compelling to use the pixel-wise MSE error as a metric to measure the performance of the model and thus resulting in maximizing the PSNR score, this loss definition has some obvious flaws for generating perceptually high-quality images. This is because the MSE based solution is optimized when it outputs the average of all possible solutions, which might be not on the HR image manifold and can be sometimes blurry, and unreal. This phenomena is illustrated in the figure below with the blue patch as the MSE based optimal solution. To solve the problem, the authors first proposed a GAN based solution to capture the natural image manifold, and a hybrid loss of summing the context loss and the adversarial loss. To further improve performance, the authors also came up with an improved context loss, which compares more high level features of the image through looking at intermediate activation of the pre-trained VGG-19 network.