EnvSDD: Benchmarking Environmental Sound Deepfake Detection

Audio generation systems now create very realistic soundscapes that can enhance media production, but also pose potential risks. Several studies have examined deepfakes in speech or singing voice. However, environmental sounds have different characteristics, which may make methods for detecting speech and singing deepfakes less effective for real-world sounds. In addition, existing datasets for environmental sound deepfake detection are limited in scale and audio types. To address this gap, we introduce EnvSDD, the first large-scale curated dataset designed for this task, consisting of 45.25 hours of real and 316.74 hours of fake audio. The test set includes diverse conditions to evaluate the generalizability, such as unseen generation models and unseen datasets. We also propose an audio deepfake detection system, based on a pre-trained audio foundation model. Results on EnvSDD show that our proposed system outperforms the state-of-the-art systems from speech and singing domains.

Towards Inclusive ASR: Investigating Voice Conversion for Dysarthric Speech Recognition in Low-Resource Languages

Automatic speech recognition (ASR) for dysarthric speech remains challenging due to data scarcity, particularly in non-English languages. To address this, we fine-tune a voice conversion model on English dysarthric speech (UASpeech) to encode both speaker characteristics and prosodic distortions, then apply it to convert healthy non-English speech (FLEURS) into non-English dysarthric-like speech. The generated data is then used to fine-tune a multilingual ASR model, Massively Multilingual Speech (MMS), for improved dysarthric speech recognition. Evaluation on PC-GITA (Spanish), EasyCall (Italian), and SSNCE (Tamil) demonstrates that VC with both speaker and prosody conversion significantly outperforms the off-the-shelf MMS performance and conventional augmentation techniques such as speed and tempo perturbation. Objective and subjective analyses of the generated data further confirm that the generated speech simulates dysarthric characteristics.

AdaKWS: Towards Robust Keyword Spotting with Test-Time Adaptation

Spoken keyword spotting (KWS) aims to identify keywords in audio for wide applications, especially on edge devices. Current small-footprint KWS systems focus on efficient model designs. However, their inference performance can decline in unseen environments or noisy backgrounds. Test-time adaptation (TTA) helps models adapt to test samples without needing the original training data. In this study, we present AdaKWS, the first TTA method for robust KWS to the best of our knowledge. Specifically, 1) We initially optimize the model's confidence by selecting reliable samples based on prediction entropy minimization and adjusting the normalization statistics in each batch. 2) We introduce pseudo-keyword consistency (PKC) to identify critical, reliable features without overfitting to noise. Our experiments show that AdaKWS outperforms other methods across various conditions, including Gaussian noise and real-scenario noises. The code will be released in due course.

Listen, Analyze, and Adapt to Learn New Attacks: An Exemplar-Free Class Incremental Learning Method for Audio Deepfake Source Tracing

As deepfake speech becomes common and hard to detect, it is vital to trace its source. Recent work on audio deepfake source tracing (ST) aims to find the origins of synthetic or manipulated speech. However, ST models must adapt to learn new deepfake attacks while retaining knowledge of the previous ones. A major challenge is catastrophic forgetting, where models lose the ability to recognize previously learned attacks. Some continual learning methods help with deepfake detection, but multi-class tasks such as ST introduce additional challenges as the number of classes grows. To address this, we propose an analytic class incremental learning method called AnaST. When new attacks appear, the feature extractor remains fixed, and the classifier is updated with a closed-form analytical solution in one epoch. This approach ensures data privacy, optimizes memory usage, and is suitable for online training. The experiments carried out in this work show that our method outperforms the baselines.

AnalyticKWS: Towards Exemplar-Free Analytic Class Incremental Learning for Small-footprint Keyword Spotting

Keyword spotting (KWS) offers a vital mechanism to identify spoken commands in voice-enabled systems, where user demands often shift, requiring models to learn new keywords continually over time. However, a major problem is catastrophic forgetting, where models lose their ability to recognize earlier keywords. Although several continual learning methods have proven their usefulness for reducing forgetting, most existing approaches depend on storing and revisiting old data to combat catastrophic forgetting. Though effective, these methods face two practical challenges: 1) privacy risks from keeping user data and 2) large memory and time consumption that limit deployment on small devices. To address these issues, we propose an exemplar-free Analytic Continual Learning (AnalyticKWS) method that updates model parameters without revisiting earlier data. Inspired by efficient learning principles, AnalyticKWS computes a closed-form analytical solution for model updates and requires only a single epoch of adaptation for incoming keywords. AnalyticKWS demands fewer computational resources by avoiding gradient-based updates and does not store old data. By eliminating the need for back-propagation during incremental learning, the model remains lightweight and efficient. As a result, AnalyticKWS meets the challenges mentioned earlier and suits resource-limited settings well. Extensive experiments on various datasets and settings show that AnalyticKWS consistently outperforms existing continual learning methods.

Nes2Net: A Lightweight Nested Architecture for Foundation Model Driven Speech Anti-spoofing

Speech foundation models have significantly advanced various speech-related tasks by providing exceptional representation capabilities. However, their high-dimensional output features often create a mismatch with downstream task models, which typically require lower-dimensional inputs. A common solution is to apply a dimensionality reduction (DR) layer, but this approach increases parameter overhead, computational costs, and risks losing valuable information. To address these issues, we propose Nested Res2Net (Nes2Net), a lightweight back-end architecture designed to directly process high-dimensional features without DR layers. The nested structure enhances multi-scale feature extraction, improves feature interaction, and preserves high-dimensional information. We first validate Nes2Net on CtrSVDD, a singing voice deepfake detection dataset, and report a 22% performance improvement and an 87% back-end computational cost reduction over the state-of-the-art baseline. Additionally, extensive testing across four diverse datasets: ASVspoof 2021, ASVspoof 5, PartialSpoof, and In-the-Wild, covering fully spoofed speech, adversarial attacks, partial spoofing, and real-world scenarios, consistently highlights Nes2Net's superior robustness and generalization capabilities. The code package and pre-trained models are available at https://github.com/Liu-Tianchi/Nes2Net.

Multi-modal Speech Enhancement with Limited Electromyography Channels

Speech enhancement (SE) aims to improve the clarity, intelligibility, and quality of speech signals for various speech enabled applications. However, air-conducted (AC) speech is highly susceptible to ambient noise, particularly in low signal-to-noise ratio (SNR) and non-stationary noise environments. Incorporating multi-modal information has shown promise in enhancing speech in such challenging scenarios. Electromyography (EMG) signals, which capture muscle activity during speech production, offer noise-resistant properties beneficial for SE in adverse conditions. Most previous EMG-based SE methods required 35 EMG channels, limiting their practicality. To address this, we propose a novel method that considers only 8-channel EMG signals with acoustic signals using a modified SEMamba network with added cross-modality modules. Our experiments demonstrate substantial improvements in speech quality and intelligibility over traditional approaches, especially in extremely low SNR settings. Notably, compared to the SE (AC) approach, our method achieves a significant PESQ gain of 0.235 under matched low SNR conditions and 0.527 under mismatched conditions, highlighting its robustness.

XLSR-Mamba: A Dual-Column Bidirectional State Space Model for Spoofing Attack Detection

Transformers and their variants have achieved great success in speech processing. However, their multi-head self-attention mechanism is computationally expensive. Therefore, one novel selective state space model, Mamba, has been proposed as an alternative. Building on its success in automatic speech recognition, we apply Mamba for spoofing attack detection. Mamba is well-suited for this task as it can capture the artifacts in spoofed speech signals by handling long-length sequences. However, Mamba's performance may suffer when it is trained with limited labeled data. To mitigate this, we propose combining a new structure of Mamba based on a dual-column architecture with self-supervised learning, using the pre-trained wav2vec 2.0 model. The experiments show that our proposed approach achieves competitive results and faster inference on the ASVspoof 2021 LA and DF datasets, and on the more challenging In-the-Wild dataset, it emerges as the strongest candidate for spoofing attack detection. The code will be publicly released in due course.

Leveraging LLM and Text-Queried Separation for Noise-Robust Sound Event Detection

Sound Event Detection (SED) is challenging in noisy environments where overlapping sounds obscure target events. Language-queried audio source separation (LASS) aims to isolate the target sound events from a noisy clip. However, this approach can fail when the exact target sound is unknown, particularly in noisy test sets, leading to reduced performance. To address this issue, we leverage the capabilities of large language models (LLMs) to analyze and summarize acoustic data. By using LLMs to identify and select specific noise types, we implement a noise augmentation method for noise-robust fine-tuning. The fine-tuned model is applied to predict clip-wise event predictions as text queries for the LASS model. Our studies demonstrate that the proposed method improves SED performance in noisy environments. This work represents an early application of LLMs in noise-robust SED and suggests a promising direction for handling overlapping events in SED. Codes and pretrained models are available at https://github.com/apple-yinhan/Noise-robust-SED.

TF-Mamba: A Time-Frequency Network for Sound Source Localization

Sound source localization (SSL) determines the position of sound sources using multi-channel audio data. It is commonly used to improve speech enhancement and separation. Extracting spatial features is crucial for SSL, especially in challenging acoustic environments. Previous studies performed well based on long short-term memory models. Recently, a novel scalable SSM referred to as Mamba demonstrated notable performance across various sequence-based modalities, including audio and speech. This study introduces the Mamba for SSL tasks. We consider the Mamba-based model to analyze spatial features from speech signals by fusing both time and frequency features, and we develop an SSL system called TF-Mamba. This system integrates time and frequency fusion, with Bidirectional Mamba managing both time-wise and frequency-wise processing. We conduct the experiments on the simulated dataset and the LOCATA dataset. Experiments show that TF-Mamba significantly outperforms other advanced methods on simulated and real-world data.