JASTIN: Aligning LLMs for Zero-Shot Audio and Speech Evaluation via Natural Language Instructions

The rapid advancement of generative audio models has outpaced the development of robust evaluation methodologies. Existing objective metrics and general multimodal large language models (MLLMs) often struggle with domain generalization, zero-shot capabilities, and instructional flexibility. To address these bottlenecks, we propose JASTIN, a generalizable, instruction-driven audio evaluation framework that formulates audio assessment as a self-instructed reasoning task. JASTIN bridges a frozen high-performance audio encoder with a fine-tuned LLM backbone via a trainable audio adapter. To ensure robust zero-shot generalization, we introduce a comprehensive instruction following data preparation pipeline, incorporating Multi-Source, Multi-Task, Multi-Calibration, and Multi-Description data. Experimental results demonstrate that JASTIN achieves state-of-the-art Pearson and Spearman correlations with human subjective ratings. It consistently outperforms general MLLMs across speech, sound, music, and out-of-domain evaluation tasks without the need for task-specific retraining.

Enhancing Conversational TTS with Cascaded Prompting and ICL-Based Online Reinforcement Learning

Conversational AI has made significant progress, yet generating expressive and controllable text-to-speech (TTS) remains challenging. Specifically, controlling fine-grained voice styles and emotions is notoriously difficult and typically requires massive amounts of heavily annotated training data. To overcome this data bottleneck, we present a scalable, data-efficient cascaded framework that pairs textual style tokens with human-curated, high-quality audio prompts. This approach enables single-shot adaptation to fine-grained speaking styles and character voices. In the context of TTS, this audio prompting acts as In-Context Learning (ICL), guiding the model's prosody and timbre without requiring massive parameter updates or large-scale retraining. To further enhance generation quality and mitigate hallucinations, we introduce a novel ICL-based online reinforcement learning (RL) strategy. This strategy directly optimizes the autoregressive prosody model using subjective aesthetic rewards while being constrained by Connectionist Temporal Classification (CTC) alignment to preserve intelligibility. Comprehensive human perception evaluations demonstrate significant improvements in both the naturalness and expressivity of the synthesized speech, establishing the efficacy of our ICL-based online RL approach.

T-Mimi: A Transformer-based Mimi Decoder for Real-Time On-Phone TTS

Neural audio codecs provide promising acoustic features for speech synthesis, with representative streaming codecs like Mimi providing high-quality acoustic features for real-time Text-to-Speech (TTS) applications. However, Mimi's decoder, which employs a hybrid transformer and convolution architecture, introduces significant latency bottlenecks on edge devices due to the the compute intensive nature of deconvolution layers which are not friendly for mobile-CPUs, such as the most representative framework XNNPACK. This paper introduces T-Mimi, a novel modification of the Mimi codec decoder that replaces its convolutional components with a purely transformer-based decoder, inspired by the TS3-Codec architecture. This change dramatically reduces on-device TTS latency from 42.1ms to just 4.4ms. Furthermore, we conduct quantization aware training and derive a crucial finding: the final two transformer layers and the concluding linear layers of the decoder, which are close to the waveform, are highly sensitive to quantization and must be preserved at full precision to maintain audio quality.

Forecasting Automotive Supply Chain Shortfalls with Heterogeneous Time Series

Operational disruptions can significantly impact companies performance. Ford, with its 37 plants globally, uses 17 billion parts annually to manufacture six million cars and trucks. With up to ten tiers of suppliers between the company and raw materials, any extended disruption in this supply chain can cause substantial financial losses. Therefore, the ability to forecast and identify such disruptions early is crucial for maintaining seamless operations. In this study, we demonstrate how we construct a dataset consisting of many multivariate time series to forecast first-tier supply chain disruptions, utilizing features related to capacity, inventory, utilization, and processing, as outlined in the classical Factory Physics framework. This dataset is technically challenging due to its vast scale of over five hundred thousand time series. Furthermore, these time series, while exhibiting certain similarities, also display heterogeneity within specific subgroups. To address these challenges, we propose a novel methodology that integrates an enhanced Attention Sequence to Sequence Deep Learning architecture, using Neural Network Embeddings to model group effects, with a Survival Analysis model. This model is designed to learn intricate heterogeneous data patterns related to operational disruptions. Our model has demonstrated a strong performance, achieving 0.85 precision and 0.8 recall during the Quality Assurance (QA) phase across Ford's five North American plants. Additionally, to address the common criticism of Machine Learning models as black boxes, we show how the SHAP framework can be used to generate feature importance from the model predictions. It offers valuable insights that can lead to actionable strategies and highlights the potential of advanced machine learning for managing and mitigating supply chain risks in the automotive industry.