Machine learning models trained with passive sensor data from mobile devices can be used to perform various inferences pertaining to activity recognition, context awareness, and health and well-being. Prior work has improved inference performance through the use of multimodal sensors (inertial, GPS, proximity, app usage, etc.) or improved machine learning. In this context, a few studies shed light on critical issues relating to the poor cross-country generalization of models due to distributional shifts across countries. However, these studies have largely relied on inference performance as a means of studying generalization issues, failing to investigate whether the root cause of the problem is linked to specific sensor modalities (independent variables) or the target attribute (dependent variable). In this paper, we study this issue in complex activities of daily living (ADL) inference task, involving 12 classes, by using a multimodal, multi-country dataset collected from 689 participants across eight countries. We first show that the ‘country of origin’ of data is captured by sensors and can be inferred from each modality separately, with an average accuracy of 65%. We then propose two diversity scores (DS) that measure how a country differentiates from others w.r.t. sensor modalities or activities. Using these diversity scores, we observed that both individual sensor modalities and activities have the ability to differentiate countries. However, while many activities capture country differences, only the ‘App usage’ and ‘Location’ sensors can do so. By dissecting country-level diversity across dependent and independent variables, we provide a framework to better understand model generalization issues across countries and country-level diversity of sensing modalities.

Natural language processing techniques are dependent upon punctuation to work well. When their input is taken from speech recognition, it is necessary to reconstruct the punctuation; in particular sentence boundaries. We define a range of features from low level acoustics to those with high level lexical semantics, including deep and recurrent models; these in turn are representative of a broad range of approaches used by previous authors for punctuation prediction. We combine the features using a gradient boosting machine that is also capable of indicating the relative importance of each feature. In an empirical study, we show that features from different semantic levels are in fact complementary, that combining statistical and deep learning methods yields better prediction results, and that generalization across different speaking styles is difficult to achieve without adaptation. Our best model achieves an F-Measure of 82.8 on a challenging broadcast news dataset.

Keyword Spotting (KWS) systems allow detecting a set of spoken (pre-defined) keywords. Open-vocabulary KWS systems search for the keywords in the set of word hypotheses generated by an automatic speech recognition (ASR) system which is computationally expensive and, therefore, often implemented as a cloud-based service. Besides, KWS systems could use also word classification algorithms that do not allow easily changing the set of words to be recognized, as the classes have to be defined a priori, even before training the system. In this paper, we propose the implementation of an open-vocabulary ASR-free KWS system based on speech and text encoders that allow matching the computed embeddings in order to spot whether a keyword has been uttered. This approach would allow choosing the set of keywords a posteriori while requiring low computational power. The experiments, performed on two different datasets, show that our method is competitive with other state of the art KWS systems while allowing for a flexibility of configuration and being computationally efficient.

Automatic evaluation of non-native speech accentedness has potential implications for not only language learning and accent identification systems but also for speaker and speech recognition systems. From the perspective of speech production, the two primary factors influencing the accentedness are the phonetic and prosodic structure. In this paper, we propose an approach for automatic accentedness evaluation based on comparison of instances of native and non-native speakers at the acoustic-phonetic level. Specifically, the proposed approach measures accentedness by comparing phone class conditional probability sequences corresponding to the instances of native and non-native speakers, respectively. We evaluate the proposed approach on the EMIME bilingual and EMIME Mandarin bilingual corpora, which contains English speech from native English speakers and various non-native English speakers, namely Finnish, German and Mandarin. We also investigate the influence of the granularity of the phonetic unit representation on the performance of the proposed accentedness measure. Our results indicate that the accentedness ratings by the proposed approach correlate consistently with the human ratings of accentedness. In addition, our studies show that the granularity of the phonetic unit representation that yields the best correlation with the human accentedness ratings varies with respect to the native language of the non-native speakers.

Manual transcription of audio databases for automatic speech recognition (ASR) training is a costly and time-consuming process. State-of-the-art hybrid ASR systems that are based on deep neural networks (DNN) can exploit un-transcribed foreign data during unsupervised DNN pre-training or semi-supervised DNN training. We investigate the relevance of foreign data characteristics, in particular domain and language. Using three different datasets of the MediaParl and Ester databases, our experiments suggest that domain and language are equally important. Foreign data recorded under matched conditions (language and domain) yields the most improvement. The resulting ASR system yields about 5% relative improvement compared to the baseline system only trained on transcribed data. Our studies also reveal that the amount of foreign data used for semi-supervised training can be significantly reduced without degrading the ASR performance if confidence measure based data selection is employed.

MediaParl is a Swiss accented bilingual database containing recordings in both French and German as they are spoken in Switzerland. The data were recorded at the Valais Parliament. Valais is a bilingual Swiss canton with many local accents and dialects. Therefore, the database contains data with high variability and is suitable to study multilingual, accented and non-native speech recognition as well as language identification and language switch detection. We also define monolingual and mixed language automatic speech recognition and language identification tasks and evaluate baseline systems. The database is publicly available for download.

The Automatic Content Linking Device monitors a conversation and uses automatically recognized words to retrieve documents that are of potential use to the participants. The document set includes project related reports or emails, transcribed snippets of past meetings, and websites. Retrieval results are displayed at regular intervals.