FastAudio: A Learnable Audio Front-End For Spoof Speech Detection

Fu, Quchen; Teng, Zhongwei; White, Jules; Powell, Maria; Schmidt, Douglas C.

doi:10.1109/icassp43922.2022.9746722

Cited by 24 publications

(6 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As test sets of ADD 2022 include unseen genuine and fake utterances which are not present in train and development data, it is essential to develop CMs that are robust to out-of-domain data. Data augmentation strategy is efficient to improve the performance of anti-spoofing systems on cross-dataset in previous works [23][24][25][26][27][28][29][30][31][32][33]. Thus, we design low-quality data augmentation strategy to address the unseen genuine and fake utterances.…”

Section: Data Augmentationmentioning

confidence: 99%

Deep Spectro-temporal Artifacts for Detecting Synthesized Speech

Liu

Zhang

et al. 2022

Proceedings of the 1st International Workshop on Deepfake Detection for Audio Multimedia

View full text Add to dashboard Cite

The Audio Deep Synthesis Detection (ADD) Challenge has been held to detect generated human-like speech. With our submitted system, this paper provides an overall assessment of track 1 (Lowquality Fake Audio Detection) and track 2 (Partially Fake Audio Detection). In this paper, spectro-temporal artifacts were detected using raw temporal signals, spectral features, as well as deep embedding features. To address track 1, low-quality data augmentation, domain adaptation via finetuning, and various complementary feature information fusion were aggregated in our system. Furthermore, we analyzed the clustering characteristics of subsystems with different features by visualization method and explained the effectiveness of our proposed greedy fusion strategy. As for track 2, frame transition and smoothing were detected using self-supervised learning structure to capture the manipulation of PF attacks in the time domain. We ranked 4th and 5th in track 1 and track 2, respectively.

show abstract

Section: Data Augmentationmentioning

confidence: 99%

Deep Spectro-temporal Artifacts for Detecting Synthesized Speech

Liu

Zhang

et al. 2022

Proceedings of the 1st International Workshop on Deepfake Detection for Audio Multimedia

View full text Add to dashboard Cite

show abstract

“…For (1), parametric filters in the frequency domain need few parameters (e.g. centre frequency, bandwidth) [8,12,13,14], but computation requires the STFT and the additional design choices of window function and framing settings (which we wish to avoid). Time-domain filterbanks have become more common [5,6,15] and are highly suitable for use with CNNs [16].…”

Section: Learnable Frontendsmentioning

confidence: 99%

Learnable Frontends that do not Learn: Quantifying Sensitivity to Filterbank Initialisation

Anderson¹,

Kinnunen²,

Harte³

2023

Preprint

View full text Add to dashboard Cite

While much of modern speech and audio processing relies on deep neural networks trained using fixed audio representations, recent studies suggest great potential in acoustic frontends learnt jointly with a backend. In this study, we focus specifically on learnable filterbanks. Prior studies have reported that in frontends using learnable filterbanks initialised to a mel scale, the learned filters do not differ substantially from their initialisation. Using a Gabor-based filterbank, we investigate the sensitivity of a learnable filterbank to its initialisation using several initialisation strategies on two audio tasks: voice activity detection and bird species identification. We use the Jensen-Shannon Distance and analysis of the learned filters before and after training. We show that although performance is overall improved, the filterbanks exhibit strong sensitivity to their initialisation strategy. The limited movement from initialised values suggests that alternate optimisation strategies may allow a learnable frontend to reach better overall performance.

show abstract

“…Speech filters are trained alongside the rest of the neural network to optimise the considered objective [28]. For instance, in FastAudio, triangular FBanks are initialized following the standard mel-scale, before adapting their central frequencies and frequency bands during training.…”

Section: Afes Inspired By Speech Processingmentioning

confidence: 99%

On the (In)Efficiency of Acoustic Feature Extractors for Self-Supervised Speech Representation Learning

Parcollet¹,

Zhang²,

Rogier³

et al. 2023

Interspeech 2023

View full text Add to dashboard Cite

Speech representations learned with self-supervised learning (SSL) have the potential to significantly improve the performance of a number of audio applications, especially when availability of labeled data from the deployment domain is limited. Despite their successes, SSL training methods are compute-and memory-heavy, and require large investments in computing infrastructure, thus putting it out of the reach of most institutions. Therefore, building efficient model architectures is essential for the wide-scale adoption of SSL in speech technologies. CNN-based Acoustic Feature Extractors (AFE), which are widely used as encoders of acoustic waveforms, remain one of the main efficiency bottlenecks. This work proposes replacing CNN-based AFEs with more efficient ones and demonstrates that SSL pre-training time and memory consumption can be reduced by a factor of two to three over existing methods while preserving performances in speech-, command-, and speakerrecognition tasks.

show abstract

FastAudio: A Learnable Audio Front-End For Spoof Speech Detection

Cited by 24 publications

References 17 publications

Deep Spectro-temporal Artifacts for Detecting Synthesized Speech

Deep Spectro-temporal Artifacts for Detecting Synthesized Speech

Learnable Frontends that do not Learn: Quantifying Sensitivity to Filterbank Initialisation

On the (In)Efficiency of Acoustic Feature Extractors for Self-Supervised Speech Representation Learning

Contact Info

Product

Resources

About