Lack of training data presents a grand challenge to scaling out spoken language understanding (SLU) to low-resource languages. Although various data augmentation approaches have been proposed to synthesize training data in low-resource target languag
es, the augmented data sets are often noisy, and thus impede the performance of SLU models. In this paper we focus on mitigating noise in augmented data. We develop a denoising training approach. Multiple models are trained with data produced by various augmented methods. Those models provide supervision signals to each other. The experimental results show that our method outperforms the existing state of the art by 3.05 and 4.24 percentage points on two benchmark datasets, respectively. The code will be made open sourced on github.
With the early success of query-answer assistants such as Alexa and Siri, research attempts to expand system capabilities of handling service automation are now abundant. However, preliminary systems have quickly found the inadequacy in relying on si
mple classification techniques to effectively accomplish the automation task. The main challenge is that the dialogue often involves complexity in user's intents (or purposes) which are multiproned, subject to spontaneous change, and difficult to track. Furthermore, public datasets have not considered these complications and the general semantic annotations are lacking which may result in zero-shot problem. Motivated by the above, we propose a Label-Aware BERT Attention Network (LABAN) for zero-shot multi-intent detection. We first encode input utterances with BERT and construct a label embedded space by considering embedded semantics in intent labels. An input utterance is then classified based on its projection weights on each intent embedding in this embedded space. We show that it successfully extends to few/zero-shot setting where part of intent labels are unseen in training data, by also taking account of semantics in these unseen intent labels. Experimental results show that our approach is capable of detecting many unseen intent labels correctly. It also achieves the state-of-the-art performance on five multi-intent datasets in normal cases.
The evaluation campaign of the International Conference on Spoken Language Translation (IWSLT 2021) featured this year four shared tasks: (i) Simultaneous speech translation, (ii) Offline speech translation, (iii) Multilingual speech translation, (iv
) Low-resource speech translation. A total of 22 teams participated in at least one of the tasks. This paper describes each shared task, data and evaluation metrics, and reports results of the received submissions.
Spoken language understanding, usually including intent detection and slot filling, is a core component to build a spoken dialog system. Recent research shows promising results by jointly learning of those two tasks based on the fact that slot fillin
g and intent detection are sharing semantic knowledge. Furthermore, attention mechanism boosts joint learning to achieve state-of-the-art results. However, current joint learning models ignore the following important facts: 1. Long-term slot context is not traced effectively, which is crucial for future slot filling. 2. Slot tagging and intent detection could be mutually rewarding, but bi-directional interaction between slot filling and intent detection remains seldom explored. In this paper, we propose a novel approach to model long-term slot context and to fully utilize the semantic correlation between slots and intents. We adopt a key-value memory network to model slot context dynamically and to track more important slot tags decoded before, which are then fed into our decoder for slot tagging. Furthermore, gated memory information is utilized to perform intent detection, mutually improving both tasks through global optimization. Experiments on benchmark ATIS and Snips datasets show that our model achieves state-of-the-art performance and outperforms other methods, especially for the slot filling task.
Abstract We introduce Generative Spoken Language Modeling, the task of learning the acoustic and linguistic characteristics of a language from raw audio (no text, no labels), and a set of metrics to automatically evaluate the learned representations
at acoustic and linguistic levels for both encoding and generation. We set up baseline systems consisting of a discrete speech encoder (returning pseudo-text units), a generative language model (trained on pseudo- text), and a speech decoder (generating a waveform from pseudo-text) all trained without supervision and validate the proposed metrics with human evaluation. Across 3 speech encoders (CPC, wav2vec 2.0, HuBERT), we find that the number of discrete units (50, 100, or 200) matters in a task-dependent and encoder- dependent way, and that some combinations approach text-based systems.1
