Electrical Engineering And Systems Science

Speech recognition is a well developed research field so that the current state of the art systems are being used in many applications in the software industry, yet as by today, there still does not exist such robust system for the recognition of words and sentences from singing voice. This paper proposes a complete pipeline for this task which may commonly be referred as automatic lyrics transcription (ALT). We have trained convolutional time-delay neural networks with self-attention on monophonic karaoke recordings using a sequence classification objective for building the acoustic model. The dataset used in this study, DAMP - Sing! 300x30x2 [1] is filtered to have songs with only English lyrics. Different language models are tested including MaxEnt and Recurrent Neural Networks based methods which are trained on the lyrics of pop songs in English. An in-depth analysis of the self-attention mechanism is held while tuning its context width and the number of attention heads. Using the best settings, our system achieves notable improvement to the state-of-the-art in ALT and provides a new baseline for the task.

Fusion of scores is a cornerstone of multimodal biometric systems composed of independent unimodal parts. In this work, we focus on quality-dependent fusion for speaker-face verification. To this end, we propose a universal model which can be trained for automatic quality assessment of both face and speaker modalities. This model estimates the quality of representations produced by unimodal systems which are then used to enhance the score-level fusion of speaker and face verification modules. We demonstrate the improvements brought by this quality-dependent fusion on the recent NIST SRE19 Audio-Visual Challenge dataset.

Phones, the segmental units of the International Phonetic Alphabet (IPA), are used for lexical distinctions in most human languages; Tones, the suprasegmental units of the IPA, are used in perhaps 70%. Many previous studies have explored cross-lingual adaptation of automatic speech recognition (ASR) phone models, but few have explored the multilingual and cross-lingual transfer of synchronization between phones and tones. In this paper, we test four Connectionist Temporal Classification (CTC)-based acoustic models, differing in the degree of synchrony they impose between phones and tones. Models are trained and tested multilingually in three languages, then adapted and tested cross-lingually in a fourth. Both synchronous and asynchronous models are effective in both multilingual and cross-lingual settings. Synchronous models achieve lower error rate in the joint phone+tone tier, but asynchronous training results in lower tone error rate.

We propose a novel architecture and improved training objectives for non-parallel voice conversion. Our proposed CycleGAN-based model performs a shape-preserving transformation directly on a high frequency-resolution magnitude spectrogram, converting its style (i.e. speaker identity) while preserving the speech content. Throughout the entire conversion process, the model does not resort to compressed intermediate representations of any sort (e.g. mel spectrogram, low resolution spectrogram, decomposed network feature). We propose an efficient axial residual block architecture to support this expensive procedure and various modifications to the CycleGAN losses to stabilize the training process. We demonstrate via experiments that our proposed model outperforms Scyclone and shows a comparable or better performance to that of CycleGAN-VC2 even without employing a neural vocoder.

This paper describes joint effort of BUT and Telefónica Research on development of Automatic Speech Recognition systems for Albayzin 2020 Challenge. We compare approaches based on either hybrid or end-to-end models. In hybrid modelling, we explore the impact of SpecAugment layer on performance. For end-to-end modelling, we used a convolutional neural network with gated linear units (GLUs). The performance of such model is also evaluated with an additional n-gram language model to improve word error rates. We further inspect source separation methods to extract speech from noisy environment (i.e. TV shows). More precisely, we assess the effect of using a neural-based music separator named Demucs. A fusion of our best systems achieved 23.33% WER in official Albayzin 2020 evaluations. Aside from techniques used in our final submitted systems, we also describe our efforts in retrieving high quality transcripts for training.

Time-domain audio separation network (TasNet) has achieved remarkable performance in blind source separation (BSS). Classic multi-channel speech processing framework employs signal estimation and beamforming. For example, Beam-TasNet links multi-channel convolutional TasNet (MC-Conv-TasNet) with minimum variance distortionless response (MVDR) beamforming, which leverages the strong modelling ability of data-driven MC-Conv-TasNet and boosts the performance of beamforming with an accurate estimation of speech statistics. Such integration can be viewed as a directed acyclic graph by accepting multi-channel input and generating multi-source output. In this letter, we design a "multi-channel input, multi-channel multi-source output" (MIMMO) speech separation system entitled "Beam-Guided TasNet", where MC-Conv-TasNet and MVDR can interact and promote each other more compactly under a directed cyclic flow. Specifically, the first stage uses Beam-TasNet to generate estimated single-speaker signals, which favours the separation in the second stage. The proposed framework facilitates iterative signal refinement with the guide of beamforming and seeks to reach the upper bound of the MVDR-based methods. Experimental results on the spatialized WSJ0-2MIX demonstrate that the Beam-Guided TasNet has achieved an SDR of 20.7 dB, which exceeded the baseline Beam-TasNet by 4.2 dB under the same model size and narrowed the gap with the oracle signal-based MVDR to 2.9 dB.

We present a bidirectional unsupervised model pre-training (UPT) method and apply it to children's automatic speech recognition (ASR). An obstacle to improving child ASR is the scarcity of child speech databases. A common approach to alleviate this problem is model pre-training using data from adult speech. Pre-training can be done using supervised (SPT) or unsupervised methods, depending on the availability of annotations. Typically, SPT performs better. In this paper, we focus on UPT to address the situations when pre-training data are unlabeled. Autoregressive predictive coding (APC), a UPT method, predicts frames from only one direction, limiting its use to uni-directional pre-training. Conventional bidirectional UPT methods, however, predict only a small portion of frames. To extend the benefits of APC to bi-directional pre-training, Bi-APC is proposed. We then use adaptation techniques to transfer knowledge learned from adult speech (using the Librispeech corpus) to child speech (OGI Kids corpus). LSTM-based hybrid systems are investigated. For the uni-LSTM structure, APC obtains similar WER improvements to SPT over the baseline. When applied to BLSTM, however, APC is not as competitive as SPT, but our proposed Bi-APC has comparable improvements to SPT.

This letter proposes a novel blind acoustic mask (BAM) designed to adaptively detect noise components and preserve target speech segments in time-domain. A robust standard deviation estimator is applied to the non-stationary noisy speech to identify noise masking elements. The main contribution of the proposed solution is the use of this noise statistics to derive an adaptive information to define and select samples with lower noise proportion. Thus, preserving speech intelligibility. Additionally, no information of the target speech and noise signals statistics is previously required to this non-ideal mask. The BAM and three competitive methods, Ideal Binary Mask (IBM), Target Binary Mask (TBM), and Non-stationary Noise Estimation for Speech Enhancement (NNESE), are evaluated considering speech signals corrupted by three non-stationary acoustic noises and six values of signal-to-noise ratio (SNR). Results demonstrate that the BAM technique achieves intelligibility gains comparable to ideal masks while maintaining good speech quality.

LPCNet is an efficient vocoder that combines linear prediction and deep neural network modules to keep the computational complexity low. In this work, we present two techniques to further reduce it's complexity, aiming for a low-cost LPCNet vocoder-based neural Text-to-Speech (TTS) System. These techniques are: 1) Sample-bunching, which allows LPCNet to generate more than one audio sample per inference; and 2) Bit-bunching, which reduces the computations in the final layer of LPCNet. With the proposed bunching techniques, LPCNet, in conjunction with a Deep Convolutional TTS (DCTTS) acoustic model, shows a 2.19x improvement over the baseline run-time when running on a mobile device, with a less than 0.1 decrease in TTS mean opinion score (MOS).

Many speech processing methods based on deep learning require an automatic and differentiable audio metric for the loss function. The DPAM approach of Manocha et al. learns a full-reference metric trained directly on human judgments, and thus correlates well with human perception. However, it requires a large number of human annotations and does not generalize well outside the range of perturbations on which it was trained. This paper introduces CDPAM, a metric that builds on and advances DPAM. The primary improvement is to combine contrastive learning and multi-dimensional representations to build robust models from limited data. In addition, we collect human judgments on triplet comparisons to improve generalization to a broader range of audio perturbations. CDPAM correlates well with human responses across nine varied datasets. We also show that adding this metric to existing speech synthesis and enhancement methods yields significant improvement, as measured by objective and subjective tests.