Intoxicated speech detection: A fusion framework with speaker-normalized hierarchical functionals and GMM supervectors
نویسندگان
چکیده
Segmental and suprasegmental speech signal modulations offer information about paralinguistic content such as affect, age and gender, pathology, and speaker state. Speaker state encompasses medium-term, temporary physiological phenomena influenced by internal or external biochemical actions (e.g., sleepiness, alcohol intoxication). Perceptual and computational research indicates that detecting speaker state from speech is a challenging task. In this paper, we present a system constructed with multiple representations of prosodic and spectral features that provided the best result at the Intoxication Subchallenge of Interspeech 2011 on the Alcohol Language Corpus. We discuss the details of each classifier and show that fusion improves performance. We additionally address the question of how best to construct a speaker state detection system in terms of robust and practical marginalization of associated variability such as through modeling speakers, utterance type, gender, and utterance length. As is the case in human perception, speaker normalization provides significant improvements to our system. We show that a held-out set of baseline (sober) data can be used to achieve comparable gains to other speaker normalization techniques. Our fused frame-level statistic-functional systems, fused GMM systems, and final combined system achieve unweighted average recalls (UARs) of 69.7%, 65.1%, and 68.8%, respectively, on the test set. More consistent numbers compared to development set results occur with matched-prompt training, where the UARs are 70.4%, 66.2%, and 71.4%, respectively. The combined system improves over the Challenge baseline by 5.5% absolute (8.4% relative), also improving upon our previously best result.
منابع مشابه
Intoxicated Speech Detection by Fusion of Speaker Normalized Hierarchical Features and GMM Supervectors
Speaker state recognition is a challenging problem due to speaker and context variability. Intoxication detection is an important area of paralinguistic speech research with potential real-world applications. In this work, we build upon a base set of various static acoustic features by proposing the combination of several different methods for this learning task. The methods include extracting ...
متن کاملSpeaker dependent emotion recognition using prosodic supervectors
This work presents a novel approach for detection of emotions embedded in the speech signal. The proposed approach works at the prosodic level, and models the statistical distribution of the prosodic features with Gaussian Mixture Models (GMM) mean-adapted from a Universal Background Model (UBM). This allows the use of GMM-mean supervectors, which are classified by a Support Vector Machine (SVM...
متن کاملSpeaker Diarization Based on Gmm Supervectors and Unsupervised Intra-speaker Variability Modeling
This paper presents a novel framework for speaker diarization. Audio is parameterized by a sequence of GMM-supervectors representing overlapping short segments of speech. Session dependent intra-session intra-speaker variability is estimated online in an unsupervised manner, and is removed from the supervectors using Nuisance Attribute Projection (NAP) The supervectors are then projected using ...
متن کاملImproving the performance of text-independent short duration SVM- and GMM-based speaker verification
In the task of automatic speaker verification (ASV) it is well known that the duration of the speech signals is an important factor in the ultimate accuracy of the system. This paper deals with some of the aspects of adapting systems to work with limited amounts of data. First we highlight the importance of a well-tuned speech detection front-end when working with short durations. We consider a...
متن کاملRead and spontaneous speech classification based on variance of GMM supervectors
This paper provides a novel method to classify spoken utterances into reading style or spontaneous style. Read/spontaneous speech classification is important for extracting data to train acoustic models for speech recognition from real data in which read speech and spontaneous speech samples are mixed. We analyzed 23,900 reading and 31,988 spontaneous utterances of 30 speakers and found that va...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
- Computer speech & language
دوره 28 2 شماره
صفحات -
تاریخ انتشار 2014