The ability to identify speech with similar emotional content is
valuable to many applications, including speech retrieval, surveillance,
and emotional speech synthesis. While current formulations
in speech emotion recognition based on classification or regression
are not appropriate for this task, solutions based on preference learning
offer appealing approaches for this task. This paper aims to
find speech samples that are emotionally similar to an anchor speech
sample provided as a query. This novel formulation opens interesting
research questions. How well can a machine complete this task?
How does the accuracy of automatic algorithms compare to the performance
of a human performing this task? This study addresses
these questions by training a deep learning model using a triplet loss
function, mapping the acoustic features into an embedding that is
discriminative for this task. The network receives an anchor speech
sample and two competing speech samples, and the task is to determine
which of the candidate speech sample conveys the closest emotional
content to the emotion conveyed by the anchor. By comparing
the results from our model with human perceptual evaluations,
this study demonstrates that the proposed approach has performance
very close to human performance in retrieving samples with similar
emotional content.
more »
« less
Preference Learning Labels by Anchoring on Consecutive Annotations
An important task in human-computer interaction is to rank speech samples according to their expressive content. A preference learning framework is appropriate for obtaining an emotional rank for a set of speech samples. However, obtaining reliable labels for training a preference learning framework is a challenging task. Most existing databases provide sentence-level absolute attribute scores annotated by multiple raters, which have to be transformed to obtain preference labels. Previous studies have shown that evaluators anchor their absolute assessments on previously annotated samples. Hence, this study proposes a novel formulation for obtaining preference learning labels by only considering annotation trends assigned by a rater to consecutive samples within an evaluation session. The experiments show that the use of the proposed anchor-based ordinal labels leads to significantly better performance than models trained using existing alternative labels.
more »
« less
- Award ID(s):
- 2016719
- NSF-PAR ID:
- 10441270
- Date Published:
- Journal Name:
- Interspeech 2023
- Page Range / eLocation ID:
- 1898 to 1902
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
to inconsistencies between annotators. The low inter-evaluator agreement arises due to the complex nature of emotions. Conventional approaches average scores provided by multiple annotators. While this approach reduces the influence of dissident annotations, previous studies have showed the value of considering individual evaluations to better capture the underlying ground-truth. One of these approaches is the qualitative agreement (QA) method, which provides an alternative framework that captures the inherent trends amongst the annotators. While previous studies have focused on using the QA method for time-continuous annotations from a fixed number of annotators, most emotional databases are annotated with attributes at the sentence-level (e.g., one global score per sentence). This study proposes a novel formulation based on the QA framework to estimate reliable sentence-level annotations for preferencelearning. The proposed relative labels between pairs of sentences capture consistent trends across evaluators. The experimental evaluation shows that preference-learning methods to rank-order emotional attributes trained with the proposed QAbased labels achieve significantly better performance than the same algorithms trained with relative scores obtained by averaging absolute scores across annotators. These results show the benefits of QA-based labels for preference-learning using sentence-level annotations.more » « less
-
Meila, Marina ; Zhang, Tong (Ed.)The label noise transition matrix, characterizing the probabilities of a training instance being wrongly annotated, is crucial to designing popular solutions to learning with noisy labels. Existing works heavily rely on finding “anchor points” or their approximates, defined as instances belonging to a particular class almost surely. Nonetheless, finding anchor points remains a non-trivial task, and the estimation accuracy is also often throttled by the number of available anchor points. In this paper, we propose an alternative option to the above task. Our main contribution is the discovery of an efficient estimation procedure based on a clusterability condition. We prove that with clusterable representations of features, using up to third-order consensuses of noisy labels among neighbor representations is sufficient to estimate a unique transition matrix. Compared with methods using anchor points, our approach uses substantially more instances and benefits from a much better sample complexity. We demonstrate the estimation accuracy and advantages of our estimates using both synthetic noisy labels (on CIFAR-10/100) and real human-level noisy labels (on Clothing1M and our self-collected human-annotated CIFAR-10). Our code and human-level noisy CIFAR-10 labels are available at https://github.com/UCSC-REAL/HOC.more » « less
-
Current leading mispronunciation detection and diagnosis (MDD) systems achieve promising performance via end-to-end phoneme recognition. One challenge of such end-to-end solutions is the scarcity of human-annotated phonemes on natural L2 speech. In this work, we leverage unlabeled L2 speech via a pseudo-labeling (PL) procedure and extend the fine-tuning approach based on pre-trained self-supervised learning (SSL) models. Specifically, we use Wav2vec 2.0 as our SSL model, and fine-tune it using original labeled L2 speech samples plus the created pseudo-labeled L2 speech samples. Our pseudo labels are dynamic and are produced by an ensemble of the online model on-the-fly, which ensures that our model is robust to pseudo label noise. We show that fine-tuning with pseudo labels achieves a 5.35% phoneme error rate reduction and 2.48% MDD F1 score improvement over a labeled-samples-only finetuning baseline. The proposed PL method is also shown to outperform conventional offline PL methods. Compared to the state-of-the-art MDD systems, our MDD solution produces a more accurate and consistent phonetic error diagnosis. In addition, we conduct an open test on a separate UTD-4Accents dataset, where our system recognition outputs show a strong correlation with human perception, based on accentedness and intelligibility.more » « less
-
NA (Ed.)This study proposes the novel formulation of measuring emotional similarity between speech recordings. This formulation explores the ordinal nature of emotions by comparing emotional similarities instead of predicting an emotional attribute, or recognizing an emotional category. The proposed task determines which of two alternative samples has the most similar emotional content to the emotion of a given anchor. This task raises some interesting questions. Which is the emotional descriptor that provide the most suitable space to assess emotional similarities? Can deep neural networks (DNNs) learn representations to robustly quantify emotional similarities? We address these questions by exploring alternative emotional spaces created with attribute-based descriptors and categorical emotions. We create the representation using a DNN trained with the triplet loss function, which relies on triplets formed with an anchor, a positive example, and a negative example. We select a positive sample that has similar emotion content to the anchor, and a negative sample that has dissimilar emotion to the anchor. The task of our DNN is to identify the positive sample. The experimental evaluations demonstrate that we can learn a meaningful embedding to assess emotional similarities, achieving higher performance than human evaluators asked to complete the same task.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.21437/Interspeech.2023-1108
