Electrical Engineering And Systems Science

State-of-the-art spoken language identification (LID) systems, which are based on end-to-end deep neural networks, have shown remarkable success not only in discriminating between distant languages but also between closely-related languages or even different spoken varieties of the same language. However, it is still unclear to what extent neural LID models generalize to speech samples with different acoustic conditions due to domain shift. In this paper, we present a set of experiments to investigate the impact of domain mismatch on the performance of neural LID systems for a subset of six Slavic languages across two domains (read speech and radio broadcast) and examine two low-level signal descriptors (spectral and cepstral features) for this task. Our experiments show that (1) out-of-domain speech samples severely hinder the performance of neural LID models, and (2) while both spectral and cepstral features show comparable performance within-domain, spectral features show more robustness under domain mismatch. Moreover, we apply unsupervised domain adaptation to minimize the discrepancy between the two domains in our study. We achieve relative accuracy improvements that range from 9% to 77% depending on the diversity of acoustic conditions in the source domain.

In this paper we describe the top-scoring IDLab submission for the text-independent task of the Short-duration Speaker Verification (SdSV) Challenge 2020. The main difficulty of the challenge exists in the large degree of varying phonetic overlap between the potentially cross-lingual trials, along with the limited availability of in-domain DeepMine Farsi training data. We introduce domain-balanced hard prototype mining to fine-tune the state-of-the-art ECAPA-TDNN x-vector based speaker embedding extractor. The sample mining technique efficiently exploits speaker distances between the speaker prototypes of the popular AAM-softmax loss function to construct challenging training batches that are balanced on the domain-level. To enhance the scoring of cross-lingual trials, we propose a language-dependent s-norm score normalization. The imposter cohort only contains data from the Farsi target-domain which simulates the enrollment data always being Farsi. In case a Gaussian-Backend language model detects the test speaker embedding to contain English, a cross-language compensation offset determined on the AAM-softmax speaker prototypes is subtracted from the maximum expected imposter mean score. A fusion of five systems with minor topological tweaks resulted in a final MinDCF and EER of 0.065 and 1.45% respectively on the SdSVC evaluation set.

Forensic audio analysis for speaker verification offers unique challenges due to location/scenario uncertainty and diversity mismatch between reference and naturalistic field recordings. The lack of real naturalistic forensic audio corpora with ground-truth speaker identity represents a major challenge in this field. It is also difficult to directly employ small-scale domain-specific data to train complex neural network architectures due to domain mismatch and loss in performance. Alternatively, cross-domain speaker verification for multiple acoustic environments is a challenging task which could advance research in audio forensics. In this study, we introduce a CRSS-Forensics audio dataset collected in multiple acoustic environments. We pre-train a CNN-based network using the VoxCeleb data, followed by an approach which fine-tunes part of the high-level network layers with clean speech from CRSS-Forensics. Based on this fine-tuned model, we align domain-specific distributions in the embedding space with the discrepancy loss and maximum mean discrepancy (MMD). This maintains effective performance on the clean set, while simultaneously generalizes the model to other acoustic domains. From the results, we demonstrate that diverse acoustic environments affect the speaker verification performance, and that our proposed approach of cross-domain adaptation can significantly improve the results in this scenario.

Developing and selecting hearing aids is a time consuming process which is simplified by using objective models. Previously, the framework for auditory discrimination experiments (FADE) accurately simulated benefits of hearing aid algorithms with root mean squared prediction errors below 3 dB. One FADE simulation requires several hours of (un)processed signals, which is obstructive when the signals have to be recorded. We propose and evaluate a data-reduced FADE version (DARF) which facilitates simulations with signals that cannot be processed digitally, but that can only be recorded in real-time. DARF simulates one speech recognition threshold (SRT) with about 30 minutes of recorded and processed signals of the (German) matrix sentence test. Benchmark experiments were carried out to compare DARF and standard FADE exhibiting small differences for stationary maskers (1 dB), but larger differences with strongly fluctuating maskers (5 dB). Hearing impairment and hearing aid algorithms seemed to reduce the differences. Hearing aid benefits were simulated in terms of speech recognition with three pairs of real hearing aids in silence ( ≥ 8 dB), in stationary and fluctuating maskers in co-located (stat. 2 dB; fluct. 6 dB), and spatially separated speech and noise signals (stat. ≥ 8 dB; fluct. 8 dB). The simulations were plausible in comparison to data from literature, but a comparison with empirical data is still open. DARF facilitates objective SRT simulations with real devices with unknown signal processing in real environments. Yet, a validation of DARF for devices with unknown signal processing is still pending since it was only tested with three similar devices. Nonetheless, DARF could be used for improving as well as for developing or model-based fitting of hearing aids.

Speech enhancement has benefited from the success of deep learning in terms of intelligibility and perceptual quality. Conventional time-frequency (TF) domain methods focus on predicting TF-masks or speech spectrum, via a naive convolution neural network (CNN) or recurrent neural network (RNN). Some recent studies use complex-valued spectrogram as a training target but train in a real-valued network, predicting the magnitude and phase component or real and imaginary part, respectively. Particularly, convolution recurrent network (CRN) integrates a convolutional encoder-decoder (CED) structure and long short-term memory (LSTM), which has been proven to be helpful for complex targets. In order to train the complex target more effectively, in this paper, we design a new network structure simulating the complex-valued operation, called Deep Complex Convolution Recurrent Network (DCCRN), where both CNN and RNN structures can handle complex-valued operation. The proposed DCCRN models are very competitive over other previous networks, either on objective or subjective metric. With only 3.7M parameters, our DCCRN models submitted to the Interspeech 2020 Deep Noise Suppression (DNS) challenge ranked first for the real-time-track and second for the non-real-time track in terms of Mean Opinion Score (MOS).

We propose a novel pitch estimation technique called DeepF0, which leverages the available annotated data to directly learns from the raw audio in a data-driven manner. F0 estimation is important in various speech processing and music information retrieval applications. Existing deep learning models for pitch estimations have relatively limited learning capabilities due to their shallow receptive field. The proposed model addresses this issue by extending the receptive field of a network by introducing the dilated convolutional blocks into the network. The dilation factor increases the network receptive field exponentially without increasing the parameters of the model exponentially. To make the training process more efficient and faster, DeepF0 is augmented with residual blocks with residual connections. Our empirical evaluation demonstrates that the proposed model outperforms the baselines in terms of raw pitch accuracy and raw chroma accuracy even using 77.4% fewer network parameters. We also show that our model can capture reasonably well pitch estimation even under the various levels of accompaniment noise.

Classic public switched telephone networks (PSTN) are often a black box for VoIP network providers, as they have no access to performance indicators, such as delay or packet loss. Only the degraded output speech signal can be used to monitor the speech quality of these networks. However, the current state-of-the-art speech quality models are not reliable enough to be used for live monitoring. One of the reasons for this is that PSTN distortions can be unique depending on the provider and country, which makes it difficult to train a model that generalizes well for different PSTN networks. In this paper, we present a new open-source PSTN speech quality test set with over 1000 crowdsourced real phone calls. Our proposed no-reference model outperforms the full-reference POLQA and no-reference P.563 on the validation and test set. Further, we analyzed the influence of file cropping on the perceived speech quality and the influence of the number of ratings and training size on the model accuracy.

The recent advances in voice conversion (VC) and text-to-speech (TTS) make it possible to produce natural sounding speech that poses threat to automatic speaker verification (ASV) systems. To this end, research on spoofing countermeasures has gained attention to protect ASV systems from such attacks. While the advanced spoofing countermeasures are able to detect known nature of spoofing attacks, they are not that effective under unknown attacks. In this work, we propose a novel data augmentation technique using a-law and mu-law based signal companding. We believe that the proposed method has an edge over traditional data augmentation by adding small perturbation or quantization noise. The studies are conducted on ASVspoof 2019 logical access corpus using light convolutional neural network based system. We find that the proposed data augmentation technique based on signal companding outperforms the state-of-the-art spoofing countermeasures showing ability to handle unknown nature of attacks.

Contrastive Predictive Coding (CPC), based on predicting future segments of speech based on past segments is emerging as a powerful algorithm for representation learning of speech signal. However, it still under-performs other methods on unsupervised evaluation benchmarks. Here, we introduce WavAugment, a time-domain data augmentation library and find that applying augmentation in the past is generally more efficient and yields better performances than other methods. We find that a combination of pitch modification, additive noise and reverberation substantially increase the performance of CPC (relative improvement of 18-22%), beating the reference Libri-light results with 600 times less data. Using an out-of-domain dataset, time-domain data augmentation can push CPC to be on par with the state of the art on the Zero Speech Benchmark 2017. We also show that time-domain data augmentation consistently improves downstream limited-supervision phoneme classification tasks by a factor of 12-15% relative.

Data efficient voice cloning aims at synthesizing target speaker's voice with only a few enrollment samples at hand. To this end, speaker adaptation and speaker encoding are two typical methods based on base model trained from multiple speakers. The former uses a small set of target speaker data to transfer the multi-speaker model to target speaker's voice through direct model update, while in the latter, only a few seconds of target speaker's audio directly goes through an extra speaker encoding model along with the multi-speaker model to synthesize target speaker's voice without model update. Nevertheless, the two methods need clean target speaker data. However, the samples provided by user may inevitably contain acoustic noise in real applications. It's still challenging to generating target voice with noisy data. In this paper, we study the data efficient voice cloning problem from noisy samples under the sequence-to-sequence based TTS paradigm. Specifically, we introduce domain adversarial training (DAT) to speaker adaptation and speaker encoding, which aims to disentangle noise from speech-noise mixture. Experiments show that for both speaker adaptation and encoding, the proposed approaches can consistently synthesize clean speech from noisy speaker samples, apparently outperforming the method adopting state-of-the-art speech enhancement module.