Transformer-based acoustic modeling has achieved great suc-cess for both hybrid and sequence-to-sequence speech recogni-tion. However, it requires access to the full sequence, and thecomputational cost grows quadratically with respect to the in-put sequence length. These factors limit its adoption for stream-ing applications. In this work, we proposed a novel augmentedmemory self-attention, which attends on a short segment of theinput sequence and a bank of memories. The memory bankstores the embedding information for all the processed seg-ments. On the librispeech benchmark, our proposed methodoutperforms all the existing streamable transformer methods bya large margin and achieved over 15% relative error reduction,compared with the widely used LC-BLSTM baseline. Our find-ings are also confirmed on some large internal datasets.
Modern wake word detection systems usually rely on neural networks for acoustic modeling. Transformers has recently shown superior performance over LSTM and convolutional networks in various sequence modeling tasks with their better temporal modeling power. However it is not clear whether this advantage still holds for short-range temporal modeling like wake word detection. Besides, the vanilla Transformer is not directly applicable to the task due to its non-streaming nature and the quadratic time and space complexity. In this paper we explore the performance of several variants of chunk-wise…
ByteDance（字节跳动）；University of California, Santa Barbara（加州大学圣巴巴拉分校）
How to find proper moments to generate partial sentence translation given a streaming speech input? Existing approaches waiting-and-translating for a fixed duration often break the acoustic units in speech, since the boundaries between acoustic units in speech are not even. In this paper, we propose MoSST, a simple yet effective method for translating streaming speech content. Given a usually long speech sequence, we develop an efficient monotonic segmentation module inside an encoder-decoder model to accumulate acoustic information incrementally and detect proper speech unit boundaries for th…
This paper presents our modeling and architecture approaches for building a highly accurate low-latency language identification system to support multilingual spoken queries for voice assistants. A common approach to solve multilingual speech recognition is to run multiple monolingual ASR systems in parallel and rely on a language identification (LID) component that detects the input language. Conventionally, LID relies on acoustic only information to detect input language. We propose an approach that learns and combines acoustic level representations with embeddings estimated on ASR hypothese…
The recurrent neural network transducer (RNN-T) is a prominent streaming end-to-end (E2E) ASR technology. In RNN-T, the acoustic encoder commonly consists of stacks of LSTMs. Very recently, as an alternative to LSTM layers, the Conformer architecture was introduced where the encoder of RNN-T is replaced with a modified Transformer encoder composed of convolutional layers at the frontend and between attention layers. In this paper, we introduce a new streaming ASR model, Convolutional Augmented Recurrent Neural Network Transducers (ConvRNN-T) in which we augment the LSTM-based RNN-T with a nove…
Institute of Acoustics, Chinese Academy of Sciences（中国科学院声学研究所）；University of Chinese Academy of Sciences（中国科学院大学）
Voice conversion models have developed for decades, and current mainstream research focuses on non-streaming voice conversion. However, streaming voice conversion is more suitable for practical application scenarios than non-streaming voice conversion. In this paper, we propose a streaming any-to-many voice conversion based on fully non-autoregressive model, which includes a streaming transformer based acoustic model and a streaming vocoder. Streaming transformer based acoustic model is composed of a pre-trained encoder from streaming end-to-end based automatic speech recognition model and a d…
StreamVoiceAnon+: Emotion-Preserving Streaming Speaker Anonymization via Frame-Level Acoustic Distillation
Nanyang Technological University, Singapore（南洋理工大学）；Institute for Infocomm Research (I2R), A*STAR, Singapore（信息与通信研究所）；The Hong Kong Polytechnic University, Hong Kong（香港理工大学）
This paper presents an end-to-end text-to-speech system with low latency on a CPU, suitable for real-time applications. The system is composed of an autoregressive attention-based sequence-to-sequence acoustic model and the LPCNet vocoder for waveform generation. An acoustic model architecture that adopts modules from both the Tacotron 1 and 2 models is proposed, while stability is ensured by using a recently proposed purely location-based attention mechanism, suitable for arbitrary sentence length generation. During inference, the decoder is unrolled and acoustic feature generation is perform…
This paper proposes an efficient memory transformer Emformer for low latency streaming speech recognition. In Emformer, the long-range history context is distilled into an augmented memory bank to reduce self-attention's computation complexity. A cache mechanism saves the computation for the key and value in self-attention for the left context. Emformer applies a parallelized block processing in training to support low latency models. We carry out experiments on benchmark LibriSpeech data. Under average latency of 960 ms, Emformer gets WER $2.50\%$ on test-clean and $5.62\%$ on test-other. Com…
We investigate training end-to-end speech recognition models with the recurrent neural network transducer (RNN-T): a streaming, all-neural, sequence-to-sequence architecture which jointly learns acoustic and language model components from transcribed acoustic data. We explore various model architectures and demonstrate how the model can be improved further if additional text or pronunciation data are available. The model consists of an `encoder', which is initialized from a connectionist temporal classification-based (CTC) acoustic model, and a `decoder' which is partially initialized from a r…
In spoken language understanding (SLU), what the user says is converted to his/her intent. Recent work on end-to-end SLU has shown that accuracy can be improved via pre-training approaches. We revisit ideas presented by Lugosch et al. using speech pre-training and three-module modeling; however, to ease construction of the end-to-end SLU model, we use as our phoneme module an open-source acoustic-phonetic model from a DNN-HMM hybrid automatic speech recognition (ASR) system instead of training one from scratch. Hence we fine-tune on speech only for the word module, and we apply multi-target le…
Conformer models have achieved state-of-the-art(SOTA) results in end-to-end speech recognition. However Conformer mainly focuses on temporal modeling while pays less attention on time-frequency property of speech feature. In this paper we augment Conformer with ConvNeXt and propose Nextformer structure. We use stacks of ConvNeXt block to replace the commonly used subsampling module in Conformer for utilizing the information contained in time-frequency speech feature. Besides, we insert an additional downsampling module in middle of Conformer layers to make our model more efficient and accurate…
We propose and evaluate transformer-based acoustic models (AMs) for hybrid speech recognition. Several modeling choices are discussed in this work, including various positional embedding methods and an iterated loss to enable training deep transformers. We also present a preliminary study of using limited right context in transformer models, which makes it possible for streaming applications. We demonstrate that on the widely used Librispeech benchmark, our transformer-based AM outperforms the best published hybrid result by 19% to 26% relative when the standard n-gram language model (LM) is u…
We present a unified and hardware efficient architecture for two stage voice trigger detection (VTD) and false trigger mitigation (FTM) tasks. Two stage VTD systems of voice assistants can get falsely activated to audio segments acoustically similar to the trigger phrase of interest. FTM systems cancel such activations by using post trigger audio context. Traditional FTM systems rely on automatic speech recognition lattices which are computationally expensive to obtain on device. We propose a streaming transformer (TF) encoder architecture, which progressively processes incoming audio chunks a…
Mispronunciation detection and diagnosis (MDD) is a popular research focus in computer-aided pronunciation training (CAPT) systems. End-to-end (e2e) approaches are becoming dominant in MDD. However an e2e MDD model usually requires entire speech utterances as input context, which leads to significant time latency especially for long paragraphs. We propose a streaming e2e MDD model called CoCA-MDD. We utilize conv-transformer structure to encode input speech in a streaming manner. A coupled cross-attention (CoCA) mechanism is proposed to integrate frame-level acoustic features with encoded refe…
The RNN-Transducers and improved attention-based encoder-decoder models are widely applied to streaming speech recognition. Compared with these two end-to-end models, the CTC model is more efficient in training and inference. However, it cannot capture the linguistic dependencies between the output tokens. Inspired by the success of two-pass end-to-end models, we introduce a transformer decoder and the two-stage inference method into the streaming CTC model. During inference, the CTC decoder first generates many candidates in a streaming fashion. Then the transformer decoder selects the best c…
Cambridge Research Laboratory, Toshiba Europe Ltd（东芝欧洲有限公司剑桥研究实验室）
In this paper, we propose an online attention mechanism, known as cumulative attention (CA), for streaming Transformer-based automatic speech recognition (ASR). Inspired by monotonic chunkwise attention (MoChA) and head-synchronous decoder-end adaptive computation steps (HS-DACS) algorithms, CA triggers the ASR outputs based on the acoustic information accumulated at each encoding timestep, where the decisions are made using a trainable device, referred to as halting selector. In CA, all the attention heads of the same decoder layer are synchronised to have a unified halting position. This fea…
Reducing prediction delay for streaming end-to-end ASR models with minimal performance regression is a challenging problem. Constrained alignment is a well-known existing approach that penalizes predicted word boundaries using external low-latency acoustic models. On the contrary, recently proposed FastEmit is a sequence-level delay regularization scheme encouraging vocabulary tokens over blanks without any reference alignments. Although all these schemes are successful in reducing delay, ASR word error rate (WER) often severely degrades after applying these delay constraining schemes. In this…
An estimated 253 million people have visual impairments. These visual impairments affect everyday lives, and limit their understanding of the outside world. This can pose a risk to health from falling or collisions. We propose a solution to this through quick and detailed communication of environmental spatial geometry through sound, providing the blind and visually impaired the ability to understand their spatial environment through sound technology. The model consists of fast object detection and 3D environmental mapping, which is communicated through a series of quick sound notes. These sou…
Neural architecture search (NAS) has attracted much attention and has been explored for automatic speech recognition (ASR). In this work, we focus on streaming ASR scenarios and propose the latency-controlled NAS for acoustic modeling. First, based on the vanilla neural architecture, normal cells are altered to causal cells to control the total latency of the architecture. Second, a revised operation space with a smaller receptive field is proposed to generate the final architecture with low latency. Extensive experiments show that: 1) Based on the proposed neural architecture, the neural netw…
Spatial audio methods are gaining a growing interest due to the spread of immersive audio experiences and applications, such as virtual and augmented reality. For these purposes, 3D audio signals are often acquired through arrays of Ambisonics microphones, each comprising four capsules that decompose the sound field in spherical harmonics. In this paper, we propose a dual quaternion representation of the spatial sound field acquired through an array of two First Order Ambisonics (FOA) microphones. The audio signals are encapsulated in a dual quaternion that leverages quaternion algebra propert…
