In this work, we present and evaluate SELMA, a Speech-Enabled Language Model for virtual Assistant interactions that integrates audio and text as inputs to a Large Language Model (LLM). SELMA is designed to handle three primary and two auxiliary tasks related to interactions with virtual assistants simultaneously within a single end-to-end model. We employ low-rank adaptation modules for parameter-efficient training of both the audio encoder and the LLM. Additionally, we implement a feature pooling strategy enabling the system to recognize global patterns and improve accuracy on tasks less reliant on individual sequence elements. Experimental results on Voice Trigger (VT) detection, Device-Directed Speech Detection (DDSD), and Automatic Speech Recognition (ASR), demonstrate that our approach both simplifies the typical input processing pipeline of virtual assistants significantly and also improves performance compared to dedicated models for each individual task. SELMA yields relative Equal-Error Rate improvements of 64% on the VT detection task, and 22% on DDSD, while also achieving word error rates close to the baseline.

Interactions with virtual assistants typically start with a predefined trigger phrase followed by the user command. To make interactions with the assistant more intuitive, we explore whether it is feasible to drop the requirement that users must begin each command with a trigger phrase. We explore this task in three ways: First, we train classifiers using only acoustic information obtained from the audio waveform. Second, we take the decoder outputs of an automatic speech recognition (ASR) system, such as 1-best hypotheses, as input features to a large language model (LLM). Finally, we explore a multimodal system that combines acoustic and lexical features, as well as ASR decoder signals in an LLM. Using multimodal information yields relative equal-error-rate improvements over text-only and audio-only models of up to 39% and 61%. Increasing the size of the LLM and training with low-rank adaption leads to further relative EER reductions of up to 18% on our dataset.

Accurate prediction of the user intent to interact with a voice assistant (VA) on a device (e.g. a smartphone) is critical for achieving naturalistic, engaging, and privacy-centric interactions with the VA. To this end, we present a novel approach to predict the user intention (whether the user is speaking to the device or not) directly from acoustic and textual information encoded at subword tokens which are obtained via an end-to-end (E2E) ASR model. Modeling directly the subword tokens, compared to modeling of the phonemes and/or full words, has at least two advantages: (i) it provides a unique vocabulary representation, where each token has a semantic meaning, in contrast to the phoneme-level representations, (ii) each subword token has a reusable "sub"-word acoustic pattern (that can be used to construct multiple full words), resulting in a largely reduced vocabulary space than of the full words. To learn the subword representations for the audio-to-intent classification, we extract: (i) acoustic information from an E2E-ASR model, which provides frame-level CTC posterior probabilities for the subword tokens, and (ii) textual information from a pretrained continuous bag-of-words model capturing the semantic meaning of the subword tokens. The key to our approach is that it combines acoustic subword-level posteriors with text information using the notion of positional-encoding to account for multiple ASR hypotheses simultaneously. We show that the proposed approach learns robust representations for audio-to-intent classification and correctly mitigates 93.3% of unintended user audio from invoking the VA at 99% true positive rate.

Suppressing unintended invocation of the device because of the speech that sounds like wake-word, or accidental button presses, is critical for a good user experience, and is referred to as False-Trigger-Mitigation (FTM). In case of multiple invocation options, the traditional approach to FTM is to use invocation-specific models, or a single model for all invocations. Both approaches are sub-optimal: the memory cost for the former approach grows linearly with the number of invocation options, which is prohibitive for on-device deployment, and does not take advantage of shared training data; while the latter is unable to accurately capture acoustic differences across different invocation types. To this end, we propose a Unified Acoustic Detector (UAD) for FTM when multiple invocation options are available on device. The proposed UAD is trained using a multi-task learning framework, where a jointly trained acoustic encoder model is augmented with invocation-specific classification layers. In the context of the FTM task, we show for the first time that using the shared model architecture across invocations (thus, keeping the model size similar to that of a monolithic model used for a single invocation type), we can not only match but largely improve the accuracy of the invocation-specific models. In particular, in the challenging case of touch-based invocation, we obtain 50% and 35% relative improvement in false positive rate at 99% true positive rate, when compared with a singleoutput model for both invocations, and separate models per invocation, respectively. Furthermore, we propose streaming and non-streaming variants of the UAD, and show that they both outperform a traditional ASR-based approach to FTM.

We address the problem of detecting speech directed to a device that does not contain a specific wake-word.Specifically, we focus on audio coming from a touch-based invocation.Mitigating virtual assistants (VAs) activation due to accidental button presses is critical for user experience.While the majority of approaches to false trigger mitigation (FTM) are designed to detect the presence of a target keyword, inferring user intent in absence of keyword is difficult.This also poses a challenge when creating the training/evaluation data for such systems due to inherent ambiguity in the user's data.To this end, we propose a novel FTM approach that uses weakly-labeled training data obtained with a newly introduced data sampling strategy.While this sampling strategy reduces data annotation efforts, the data labels are noisy as the data are not annotated manually.We use these data to train an acoustics-only model for the FTM task by regularizing its loss function via knowledge distillation from an ASR-based (LatticeRNN) model.This improves the model decisions, resulting in 66% gain in accuracy, as measured by equal-error-rate (EER), over the base acoustics-only model.We also show that the ensemble of the LatticeRNN and acousticdistilled models brings further accuracy improvement of 20%.

Voice trigger detection is an important task, which enables activating a voice assistant when a target user speaks a keyword phrase.A detector is typically trained on speech data independent of speaker information and used for the voice trigger detection task.However, such a speaker independent voice trigger detector typically suffers from performance degradation on speech from underrepresented groups, such as accented speakers.In this work, we propose a novel voice trigger detector that can use a small number of utterances from a target speaker to improve detection accuracy.Our proposed model employs an encoder-decoder architecture.While the encoder performs speaker independent voice trigger detection, similar to the conventional detector, the decoder is trained with metric learning and predicts a personalized embedding for each utterance.A personalized voice trigger score is then obtained as a similarity score between the embeddings of enrollment utterances and a test utterance.The personalized embedding allows adapting to target speaker's speech when computing the voice trigger score, hence improving voice trigger detection accuracy.Experimental results show that the proposed approach achieves a 38% relative reduction in a false rejection rate (FRR) compared to a baseline speaker independent voice trigger model.

We present an introspection of an audiovisual speech enhancement model. In particular, we focus on interpreting how a neural audiovisual speech enhancement model uses visual cues to improve the quality of the target speech signal. We show that visual cues provide not only high-level information about speech activity, i.e., speech/silence, but also fine-grained visual information about the place of articulation. One byproduct of this finding is that the learned visual embeddings can be used as features for other visual speech applications. We demonstrate the effectiveness of the learned visual embeddings for classifying visemes (the visual analogy to phonemes). Our results provide insight into important aspects of audiovisual speech enhancement and demonstrate how such models can be used for self-supervision tasks for visual speech applications.

In this paper, we address the task of determining whether a given utterance is directed towards a voice-enabled smart-assistant device or not. An undirected utterance is termed as a "false trigger" and false trigger mitigation (FTM) is essential for designing a privacy-centric non-intrusive smart assistant. The directedness of an utterance can be identified by running automatic speech recognition (ASR) and determining the user intent by analyzing the ASR transcript. Yet, in case of a false trigger, transcribing the audio using ASR itself is strongly undesirable. To alleviate this issue, we propose an LSTM-based FTM architecture which determines the user intent from acoustic features directly without explicitly generating ASR transcripts from the audio. The proposed models are small-footprint and can be run on-device with limited computational resources. During training, the model parameters are optimized using a knowledge transfer approach where a more accurate self-attention graph neural network model [1] serves as the teacher. Given the whole audio snippets, our approach mitigates 87% of false triggers at 99% true positive rate (TPR), and in a streaming audio scenario, the system listens to only 1.69s of the false trigger audio before rejecting it while achieving the same TPR.

We present an architecture for voice trigger detection for virtual assistants. The main idea in this work is to exploit information in words that immediately follow the trigger phrase. We first demonstrate that by including more audio context after a detected trigger phrase, we can indeed get a more accurate decision. However, waiting to listen to more audio each time incurs a latency increase. Progressive Voice Trigger Detection allows us to trade-off latency and accuracy by accepting clear trigger candidates quickly, but waiting for more context to decide whether to accept more marginal examples. Using a two-stage architecture, we show that by delaying the decision for just 3% of detected true triggers in the test set, we are able to obtain a relative improvement of 66% in false rejection rate, while incurring only a negligible increase in latency.

It is desirable for a text-to-speech system to take into account the environment where synthetic speech is presented, and provide appropriate context-dependent output to the user. In this paper, we present and compare various approaches for generating different speaking styles, namely, normal, Lombard, and whisper speech, using only limited data. The following systems are proposed and assessed: 1) Pre-training and fine-tuning a model for each style. 2) Lombard and whisper speech conversion through a signal processing based approach. 3) Multi-style generation using a single model based on a speaker verification model. Our mean opinion score and AB preference listening tests show that 1) we can generate high quality speech through the pre-training/fine-tuning approach for all speaking styles. 2) Although our speaker verification (SV) model is not explicitly trained to discriminate different speaking styles, and no Lombard and whisper voice is used for pretrain this system, SV model can be used as style encoder for generating different style embeddings as input for Tacotron system. We also show that the resulting synthetic Lombard speech has a significant positive impact on intelligibility gain.

Emotion plays an essential role in human-to-human communication, enabling us to convey feelings such as happiness, frustration, and sincerity. While modern speech technologies rely heavily on speech recognition and natural language understanding for speech content understanding, the investigation of vocal expression is increasingly gaining attention. Key considerations for building robust emotion models include characterizing and improving the extent to which a model, given its training data distribution, is able to generalize to unseen data conditions. This work investigated a long-shot-term memory (LSTM) network and a time convolution - LSTM (TC-LSTM) to detect primitive emotion attributes such as valence, arousal, and dominance, from speech. It was observed that training with multiple datasets and using robust features improved the concordance correlation coefficient (CCC) for valence, by 30% with respect to the baseline system. Additionally, this work investigated how emotion primitives can be used to detect categorical emotions such as happiness, disgust, contempt, anger, and surprise from neutral speech, and results indicated that arousal, followed by dominance was a better detector of such emotions.

We present progress towards bilingual Text-to-Speech which is able to transform a monolingual voice to speak a second language while preserving speaker voice quality. We demonstrate that a bilingual speaker embedding space contains a separate distribution for each language and that a simple transform in speaker space generated by the speaker embedding can be used to control the degree of accent of a synthetic voice in a language. The same transform can be applied even to monolingual speakers.In our experiments speaker data from an English-Spanish (Mexican) bilingual speaker was used, and the goal was to enable English speakers to speak Spanish and Spanish speakers to speak English. We found that the simple transform was sufficient to convert a voice from one language to the other with a high degree of naturalness. In one case the transformed voice outperformed a native language voice in listening tests. Experiments further indicated that the transform preserved many of the characteristics of the original voice. The degree of accent present can be controlled and naturalness is relatively consistent across a range of accent values.

Automatic speech transcription and speaker recognition are usually treated as separate tasks even though they are interdependent. In this study, we investigate training a single network to perform both tasks jointly. We train the network in a supervised multi-task learning setup, where the speech transcription branch of the network is trained to minimise a phonetic connectionist temporal classification (CTC) loss while the speaker recognition branch of the network is trained to label the input sequence with the correct label for the speaker. We present a large-scale empirical study where the model is trained using several thousand hours of labelled training data for each task. We evaluate the speech transcription branch of the network on a voice trigger detection task while the speaker recognition branch is evaluated on a speaker verification task. Results demonstrate that the network is able to encode both phonetic and speaker information in its learnt representations while yielding accuracies at least as good as the baseline models for each task, with the same number of parameters as the independent models.

Millions of people reach out to digital assistants such as Siri every day, asking for information, making phone calls, seeking assistance, and much more. The expectation is that such assistants should understand the intent of the users query. Detecting the intent of a query from a short, isolated utterance is a difficult task. Intent cannot always be obtained from speech-recognized transcriptions. A transcription driven approach can interpret what has been said but fails to acknowledge how it has been said, and as a consequence, may ignore the expression present in the voice. Our work investigates whether a system can reliably detect vocal expression in queries using acoustic and paralinguistic embedding. Results show that the proposed method offers a relative equal error rate (EER) decrease of 60% compared to a bag-of-word based system, corroborating that expression is significantly represented by vocal attributes, rather than being purely lexical. Addition of emotion embedding helped to reduce the EER by 30% relative to the acoustic embedding, demonstrating the relevance of emotion in expressive voice.

“Serious games” are becoming extremely relevant to individuals who have specific needs, such as children with an autism spectrum condition (ASC). Often, individuals with an ASC have difficulties in interpreting verbal and nonverbal communication cues during social interactions. The ASC-Inclusion EU-FP7 funded project aims to provide children who have an ASC with a platform to learn emotion expression and recognition, through play in the virtual world. In particular, the ASC-Inclusion platform focuses on the expression of emotion via facial, vocal, and bodily gestures. The platform combines multiple analysis tools, using onboard microphone and webcam capabilities. The platform utilizes these capabilities via training games, text-based communication, animations, video, and audio clips. This paper introduces current findings and evaluations of the ASC-Inclusion platform and provides detailed description for the different modalities.

In this paper we introduce a speaker verification system deployed on mobile devices that can be used to personalise a keyword spotter. We describe a baseline DNN system that maps an utterance to a speaker embedding, which is used to measure speaker differences via cosine similarity. We then introduce an architectural modification which uses an LSTM system where the parameters are optimised via a curriculum learning procedure to reduce the detection error and improve its generalisability across various conditions. Experiments on our internal datasets show that the proposed approach outperforms the DNN baseline system and yields a relative EER reduction of 30-70% on both text-dependent and text-independent tasks under a variety of acoustic conditions.

Dimensional emotion recognition from physiological signals is a highly challenging task. Common methods rely on hand-crafted features that do not yet provide the performance necessary for real-life application. In this work, we exploit a series of convolutional and recurrent neural networks to predict affect from physiological signals, such as electrocardiogram and electrodermal activity, directly from the raw time representation. The motivation behind this so-called end-to-end approach is that, ultimately, the network learns an intermediate representation of the physiological signals that better suits the task at hand. Experimental evaluations show that, this very first study on end-to-end learning of emotion based on physiology, yields significantly better performance in comparison to existing work on the challenging RECOLA database, which includes fully spontaneous affective behaviors displayed during naturalistic interactions. Furthermore, we gain better understanding of the models' inner representations, by demonstrating that some cells' activations in the convolutional network are correlated to a large extent with hand-crafted features.

The automatic recognition of spontaneous emotions from speech is a challenging task. On the one hand, acoustic features need to be robust enough to capture the emotional content for various styles of speaking, and while on the other, machine learning algorithms need to be insensitive to outliers while being able to model the context. Whereas the latter has been tackled by the use of Long Short-Term Memory (LSTM) networks, the former is still under very active investigations, even though more than a decade of research has provided a large set of acoustic descriptors. In this paper, we propose a solution to the problem of ‘context-aware’ emotional relevant feature extraction, by combining Convolutional Neural Networks (CNNs) with LSTM networks, in order to automatically learn the best representation of the speech signal directly from the raw time representation. In this novel work on the so-called end-to-end speech emotion recognition, we show that the use of the proposed topology significantly outperforms the traditional approaches based on signal processing techniques for the prediction of spontaneous and natural emotions on the RECOLA database.

Semi-Supervised Learning (SSL) techniques have found many applications where labeled data is scarce and/or expensive to obtain. However, SSL suffers from various inherent limitations that limit its performance in practical applications. A central problem is that the low performance that a classifier can deliver on challenging recognition tasks reduces the trustability of the automatically labeled data. Another related issue is the noise accumulation problem - instances that are misclassified by the system are still used to train it in future iterations. In this paper, we propose to address both issues in the context of emotion recognition. Initially, we exploit the complementarity between audio-visual features to improve the performance of the classifier during the supervised phase. Then, we iteratively re-evaluate the automatically labeled instances to correct possibly mislabeled data and this enhances the overall confidence of the system's predictions. Experimental results performed on the RECOLA database demonstrate that our methodology delivers a strong performance in the classification of high/low emotional arousal (UAR = 76.5%), and significantly outperforms traditional SSL methods by at least 5.0% (absolute gain).

We introduce a recurrent neural network architecture for automated road surface wetness detection from audio of tire-surface interaction. The robustness of our approach is evaluated on 785,826 bins of audio that span an extensive range of vehicle speeds, noises from the environment, road surface types, and pavement conditions including international roughness index (IRI) values from 25 in/mi to 1400 in/mi. The training and evaluation of the model are performed on different roads to minimize the impact of environmental and other external factors on the accuracy of the classification. We achieve an unweighted average recall (UAR) of 93.2% across all vehicle speeds including 0 mph. The classifier still works at 0 mph because the discriminating signal is present in the sound of other vehicles driving by.

In this paper, we propose a method for automatically detecting deceptive speech by relying on predicted scores derived from emotion dimensions such as arousal, valence, regulation, and emotion categories.The scores are derived from task-dependent models trained on the GEMEP emotional speech database.Inputs from the INTERSPEECH 2016 Computational Paralinguistics Deception sub-challenge are processed to obtain predictions of emotion attributes and associated scores that are then used as features in detecting deception.We show that using the new emotion-related features, it is possible to improve upon the challenge baseline.

International audience

We investigate, for the first time, if applying model selection based on automatic language identification (LID) can improve multilingual recognition of emotion in speech.Six emotional speech corpora from three language families (Germanic, Romance, Sino-Tibetan) are evaluated.The emotions are represented by the quadrants in the arousal/valence plane, i. e., positive/negative arousal/valence.Four selection approaches for choosing an optimal training set depending on the current language are compared: within the same language family, across language family, use of all available corpora, and selection based on the automatic LID.We found that, on average, the proposed LID approach for selecting training corpora is superior to using all the available corpora when the spoken language is not known.

International audience

In the context of the development of Serious Games, this position paper sheds light on their possible use to support and enhance social inclusion through the presentation of three research projects in this domain. We first describe the context of our works and research. We then report on several events joining the scientific community that focused on the research on Digital Games for Empowerment and Inclusion. Last, we propose some results of our experience in this domain. The presented work particularly highlights the importance of user-centered design in interdisciplinary cooperation with psychologists, sociologists, vulnerable end users and the practitioners. We stress the relevance of progress beyond the state-of-the-art in recently focused areas of AI, such as, in computational theory-of-mind, in affective multimodal computing, in learning of socio-emotional skills, and in recovery of human attention processes. Typical examples are given to illustrate the projects' research results.

We demonstrate audEERING's sensAI technology running natively on low-resource mobile devices applied to emotion analytics and speaker characterisation tasks. A showcase application for the Android platform is provided, where au-dEERING's highly noise robust voice activity detection based on Long Short-Term Memory Recurrent Neural Networks (LSTM-RNN) is combined with our core emotion recognition and speaker characterisation engine natively on the mobile device. This eliminates the need for network connectivity and allows to perform robust speaker state and trait recognition efficiently in real-time without network transmission lags. Real-time factors are benchmarked for a popular mobile device to demonstrate the efficiency, and average response times are compared to a server based approach. The output of the emotion analysis is visualized graphically in the arousal and valence space alongside the emotion category and further speaker characteristics.

Acoustic novelty detection aims at identifying abnormal/novel acoustic signals which differ from the reference/normal data that the system was trained with. In this paper we present a novel unsupervised approach based on a denoising autoencoder. In our approach auditory spectral features are processed by a denoising autoencoder with bidirectional Long Short-Term Memory recurrent neural networks. We use the reconstruction error between the input and the output of the autoencoder as activation signal to detect novel events. The autoencoder is trained on a public database which contains recordings of typical in-home situations such as talking, watching television, playing and eating. The evaluation was performed on more than 260 different abnormal events. We compare results with state-of-the-art methods and we conclude that our novel approach significantly outperforms existing methods by achieving up to 93.4% F-Measure.

Acoustic novelty detection aims at identifying abnormal/novel acoustic signals which differ from the reference/normal data that the system was trained with. In this paper we present a novel approach based on non-linear predictive denoising autoencoders. In our approach, auditory spectral features of the next short-term frame are predicted from the previous frames by means of Long-Short Term Memory (LSTM) recurrent denoising autoencoders. We show that this yields an effective generative model for audio. The reconstruction error between the input and the output of the autoencoder is used as activation signal to detect novel events. The autoencoder is trained on a public database which contains recordings of typical in-home situations such as talking, watching television, playing and eating. The evaluation was performed on more than 260 different abnormal events. We compare results with state-of-the-art methods and we conclude that our novel approach significantly outperforms existing methods by achieving up to 94.4% F-Measure.

Only a few studies exist on automatic emotion analysis of speech from children with Autism Spectrum Conditions (ASC). Out of these, some preliminary studies have recently focused on comparing the relevance of selected prosodic features against large sets of acoustic, spectral, and cepstral features; however, no study so far provided a comparison of performances across dif-ferent languages. The present contribution aims to fill this white spot in the literature and provide insight by extensive evaluations carried out on three databases of prompted phrases collected in English, Swedish, and Hebrew, inducing nine emotion categories embedded in short-stories. The datasets contain speech of chil-dren with ASC and typically developing children under the same conditions. We evaluate automatic diagnosis and recognition of emotions in atypical childrens voice over the nine categories including binary valence/arousal discrimination.

The automatic recognition of facial behaviours is usually achieved through the detection of particular FACS Action Unit (AU), which then makes it possible to analyse the affective behaviours expressed in the face.Despite the fact that advanced techniques have been proposed to extract relevant facial descriptors, the processing of real-life data, i. e., recorded in unconstrained environments, makes the automatic detection of FACS AU much more challenging compared to constrained recordings, such as posed faces, and even impossible when the corresponding parts of the face are masked or subject to low or no illumination.We present in this paper the very first attempt in using acoustic cues for the automatic detection of FACS AU, as an alternative way to obtain information from the face when such data are not available.Results show that features extracted from the voice can be effectively used to predict different types of FACS AU, and that the best performance are obtained for the prediction of the apex, in comparison to the prediction of onset, offset and occurrence.

A plethora of different onset detection methods have been proposed in the recent years. However, few attempts have been made with respect to widely-applicable approaches in order to achieve superior performances over different types of music and with considerable temporal precision. In this paper, we present a multi-resolution approach based on discrete wavelet transform and linear prediction filtering that improves time resolution and performance of onset detection in different musical scenarios. In our approach, wavelet coefficients and forward prediction errors are combined with auditory spectral features and then processed by a bidirectional Long Short-Term Memory recurrent neural network, which acts as reduction function. The network is trained with a large database of onset data covering various genres and onset types. We compare results with state-of-the-art methods on a dataset that includes Bello, Glover and ISMIR 2004 Ballroom sets, and we conclude that our approach significantly outperforms existing methods in terms of F-Measure. For pitched non percussive music an absolute improvement of 7.5% is reported.

This paper concerns the exploitation of multi-resolution time-frequency features via Wavelet Packet Transform to improve audio onset detection. In our approach, Wavelet Packet Energy Coefficients (WPEC) and Auditory Spectral Features (ASF) are processed by Bidirectional Long Short-Term Memory (BLSTM) recurrent neural network that yields the onsets location. The combination of the two feature sets, together with the BLSTM based detector, form an advanced energy-based approach that takes advantage from the multi-resolution analysis given by the wavelet decomposition of the audio input signal. The neural network is trained with a large database of onset data covering various genres and onset types. Due to its data-driven nature, our approach does not require the onset detection method and its parameters to be tuned to a particular type of music. We show a comparison with other types and sizes of recurrent neural networks and we compare results with state-of-the-art methods on the whole onset dataset. We conclude that our approach significantly increase performance in terms of F-measure without any music genres or onset type constraints.

The INTERSPEECH 2014 Computational Paralinguistics Challenge provides for the first time a unified test-bed for the automatic recognition of speakers’ cognitive and physical load in speech. In this paper, we describe these two Sub-Challenges, their conditions, baseline results and experimental procedures, as well as the COMPARE baseline features generated with the openSMILE toolkit and provided to the participants in the Challenge.

In speech emotion recognition, training and test data used for system development usually tend to fit each other perfectly, but further 'similar' data may be available. Transfer learning helps to exploit such similar data for training despite the inherent dissimilarities in order to boost a recogniser's performance. In this context, this paper presents a sparse auto encoder method for feature transfer learning for speech emotion recognition. In our proposed method, a common emotion-specific mapping rule is learnt from a small set of labelled data in a target domain. Then, newly reconstructed data are obtained by applying this rule on the emotion-specific data in a different domain. The experimental results evaluated on six standard databases show that our approach significantly improves the performance relative to learning each source domain independently.

International audience

Speech data is in principle available in large amounts for the training of acoustic emotion recognisers.However, emotional labelling is usually not given and the distribution is heavily unbalanced, as most data is 'rather neutral' than truly 'emotional'.In the 'hay stack' of speech data, Active Learning automatically identifies the 'needles', i.e., the more informative instances to reduce human labelling effort when building a classifier, e.g., for acoustic emotion recognition.The critical issue thus is the determination and quantification of informativeness.To this end, we suggest to exploit the reliability of the usual ambiguity of emotional labels, i.e., we propose a novel approach based on label uncertainty.By building a certainty model and predicting the candidate instances, informativeness is thus based on labeller agreement.In addition, we consider class sparseness.The results of extensive test runs under well standardised conditions show the method's great potential in reducing labelling costs while boosting performance.

We address the fully automatic recognition of intoxication, sleepiness, age and gender from speech in medium-term observation intervals of up to several minutes.The nature of these speaker states and traits as being medium-term or long-term, as opposed to short-term states such as emotion, makes it possible to collect cumulative evidence in the form of utterance level decisions; we show that by fusing these decisions along the time axis, more and more accurate decisions can be obtained.In extensive test runs on three official INTERSPEECH Challenge corpora, we show that the average recall can be improved by up to 5 %, 6 %, 10 % and 11 % absolute by longer-term observation of speaker sleepiness, gender, intoxication, and age, respectively, compared to the accuracy of a decision from a single utterance.