Electrical Engineering And Systems Science

Generative Models for Audio Synthesis have been gaining momentum in the last few years. More recently, parametric representations of the audio signal have been incorporated to facilitate better musical control of the synthesized output. In this work, we investigate a parametric model for violin tones, in particular the generative modeling of the residual bow noise to make for more natural tone quality. To aid in our analysis, we introduce a dataset of Carnatic Violin Recordings where bow noise is an integral part of the playing style of higher pitched notes in specific gestural contexts. We obtain insights about each of the harmonic and residual components of the signal, as well as their interdependence, via observations on the latent space derived in the course of variational encoding of the spectral envelopes of the sustained sounds.

We consider the design of two-pass voice trigger detection systems. We focus on the networks in the second pass that are used to re-score candidate segments obtained from the first-pass. Our baseline is an acoustic model(AM), with BiLSTM layers, trained by minimizing the CTC loss. We replace the BiLSTM layers with self-attention layers. Results on internal evaluation sets show that self-attention networks yield better accuracy while requiring fewer parameters. We add an auto-regressive decoder network on top of the self-attention layers and jointly minimize the CTC loss on the encoder and the cross-entropy loss on the decoder. This design yields further improvements over the baseline. We retrain all the models above in a multi-task learning(MTL) setting, where one branch of a shared network is trained as an AM, while the second branch classifies the whole sequence to be true-trigger or not. Results demonstrate that networks with self-attention layers yield ∼ 60% relative reduction in false reject rates for a given false-alarm rate, while requiring 10% fewer parameters. When trained in the MTL setup, self-attention networks yield further accuracy improvements. On-device measurements show that we observe 70% relative reduction in inference time. Additionally, the proposed network architectures are ∼ 5X faster to train.

Automatic speech recognition (ASR) is a relevant area in multiple settings because it provides a natural communication mechanism between applications and users. ASRs often fail in environments that use language specific to particular application domains. Some strategies have been explored to reduce errors in closed ASRs through post-processing, particularly automatic spell checking, and deep learning approaches. In this article, we explore using a deep neural network to refine the results of a phonetic correction algorithm applied to a telesales audio database. The results exhibit a reduction in the word error rate (WER), both in the original transcription and in the phonetic correction, which shows the viability of deep learning models together with post-processing correction strategies to reduce errors made by closed ASRs in specific language domains.

The ICASSP 2021 Acoustic Echo Cancellation Challenge is intended to stimulate research in the area of acoustic echo cancellation (AEC), which is an important part of speech enhancement and still a top issue in audio communication and conferencing systems. Many recent AEC studies report good performance on synthetic datasets where the train and test samples come from the same underlying distribution. However, the AEC performance often degrades significantly on real recordings. Also, most of the conventional objective metrics such as echo return loss enhancement (ERLE) and perceptual evaluation of speech quality (PESQ) do not correlate well with subjective speech quality tests in the presence of background noise and reverberation found in realistic environments. In this challenge, we open source two large datasets to train AEC models under both single talk and double talk scenarios. These datasets consist of recordings from more than 2,500 real audio devices and human speakers in real environments, as well as a synthetic dataset. We open source two large test sets, and we open source an online subjective test framework for researchers to quickly test their results. The winners of this challenge will be selected based on the average Mean Opinion Score (MOS) achieved across all different single talk and double talk scenarios.

The Deep Noise Suppression (DNS) challenge is designed to foster innovation in the area of noise suppression to achieve superior perceptual speech quality. We recently organized a DNS challenge special session at INTERSPEECH 2020. We open sourced training and test datasets for researchers to train their noise suppression models. We also open sourced a subjective evaluation framework and used the tool to evaluate and pick the final winners. Many researchers from academia and industry made significant contributions to push the field forward. We also learned that as a research community, we still have a long way to go in achieving excellent speech quality in challenging noisy real-time conditions. In this challenge, we are expanding both our training and test datasets. There are two tracks with one focusing on real-time denoising and the other focusing on real-time personalized deep noise suppression. We also make a non-intrusive objective speech quality metric called DNSMOS available for participants to use during their development stages. The final evaluation will be based on subjective tests.

ICE-Talk is an open source web-based GUI that allows the use of a TTS system with controllable parameters via a text field and a clickable 2D plot. It enables the study of latent spaces for controllable TTS. Moreover it is implemented as a module that can be used as part of a Human-Agent interaction.

In this paper, we introduce ID-Conditioned Auto-Encoder for unsupervised anomaly detection. Our method is an adaptation of the Class-Conditioned Auto-Encoder (C2AE) designed for the open-set recognition. Assuming that non-anomalous samples constitute of distinct IDs, we apply Conditioned Auto-Encoder with labels provided by these IDs. Opposed to C2AE, our approach omits the classification subtask and reduces the learning process to the single run. We simplify the learning process further by fixing a constant vector as the target for non-matching labels. We apply our method in the context of sounds for machine condition monitoring. We evaluate our method on the ToyADMOS and MIMII datasets from the DCASE 2020 Challenge Task 2. We conduct an ablation study to indicate which steps of our method influences results the most.

Numerous studies have investigated the effectiveness of audio-visual multimodal learning for speech enhancement (AVSE) tasks, seeking a solution that uses visual data as auxiliary and complementary input to reduce the noise of noisy speech signals. Recently, we proposed a lite audio-visual speech enhancement (LAVSE) algorithm. Compared to conventional AVSE systems, LAVSE requires less online computation and moderately solves the user privacy problem on facial data. In this study, we extend LAVSE to improve its ability to address three practical issues often encountered in implementing AVSE systems, namely, the requirement for additional visual data, audio-visual asynchronization, and low-quality visual data. The proposed system is termed improved LAVSE (iLAVSE), which uses a convolutional recurrent neural network architecture as the core AVSE model. We evaluate iLAVSE on the Taiwan Mandarin speech with video dataset. Experimental results confirm that compared to conventional AVSE systems, iLAVSE can effectively overcome the aforementioned three practical issues and can improve enhancement performance. The results also confirm that iLAVSE is suitable for real-world scenarios, where high-quality audio-visual sensors may not always be available.

This paper proposes a spectral-domain perceptual weighting technique for Parallel WaveGAN-based text-to-speech (TTS) systems. The recently proposed Parallel WaveGAN vocoder successfully generates waveform sequences using a fast non-autoregressive WaveNet model. By employing multi-resolution short-time Fourier transform (MR-STFT) criteria with a generative adversarial network, the light-weight convolutional networks can be effectively trained without any distillation process. To further improve the vocoding performance, we propose the application of frequency-dependent weighting to the MR-STFT loss function. The proposed method penalizes perceptually-sensitive errors in the frequency domain; thus, the model is optimized toward reducing auditory noise in the synthesized speech. Subjective listening test results demonstrate that our proposed method achieves 4.21 and 4.26 TTS mean opinion scores for female and male Korean speakers, respectively.

User interactions with personal assistants like Alexa, Google Home and Siri are typically initiated by a wake term or wakeword. Several personal assistants feature "follow-up" modes that allow users to make additional interactions without the need of a wakeword. For the system to only respond when appropriate, and to ignore speech not intended for it, utterances must be classified as device-directed or non-device-directed. State-of-the-art systems have largely used acoustic features for this task, while others have used only lexical features or have added LM-based lexical features. We propose a directedness classifier that combines semantic lexical features with a lightweight acoustic feature and show it is effective in classifying directedness. The mixed-domain lexical and acoustic feature model is able to achieve 14% relative reduction of EER over a state-of-the-art acoustic-only baseline model. Finally, we successfully apply transfer learning and semi-supervised learning to the model to improve accuracy even further.