Mark Huckvale

Computer-assisted therapy for medication-resistant auditory hallucinations: proof-of-concept study

BACKGROUND One in four patients with schizophrenia responds poorly to antipsychotic medication, continuing to hear persecutory auditory hallucinations. Patients who are able to sustain a dialogue with their persecutor feel much more in control. AIMS To develop a computerised system that enables the patient to create an avatar of their persecutor. To encourage them to engage in a dialogue with the avatar, which the therapist is able to control so that the avatar progressively yields control to the patient. METHOD Avatar therapy was evaluated by a randomised, single blind, partial crossover trial comparing the novel therapy with treatment as usual (TAU). We used three main outcome measures: (a) the Psychotic Symptom Rating Scale (PSYRATS), hallucinations section; (b) the Omnipotence and Malevolence subscales of the Revised Beliefs About Voices Questionnaire (BAVQ-R); and (c) the Calgary Depression Scale (CDS). RESULTS The control group showed no change over time in their scores on the three assessments, whereas the novel therapy group showed mean reductions in the total PSYRATS score (auditory hallucinations) of 8.75 (P = 0.003) and in the BAVQ-R combined score of omnipotence and malevolence of the voices of 5.88 (P = 0.004). There was no significant reduction in the CDS total score for depression. For the crossover control group, comparison of the period of TAU with the period of avatar therapy confirmed the findings of the previous analysis. The effect size of the therapy was 0.8. CONCLUSIONS Avatar therapy represents a promising treatment for medication-resistant auditory hallucinations. Replication with a larger sample is required before roll-out to clinical settings.

Avatar therapy for persecutory auditory hallucinations: What is it and how does it work?

We have developed a novel therapy based on a computer program, which enables the patient to create an avatar of the entity, human or non-human, which they believe is persecuting them. The therapist encourages the patient to enter into a dialogue with their avatar, and is able to use the program to change the avatar so that it comes under the patient’s control over the course of six 30-min sessions and alters from being abusive to becoming friendly and supportive. The therapy was evaluated in a randomised controlled trial with a partial crossover design. One group went straight into the therapy arm: “immediate therapy”. The other continued with standard clinical care for 7 weeks then crossed over into Avatar therapy: “delayed therapy”. There was a significant reduction in the frequency and intensity of the voices and in their omnipotence and malevolence. Several individuals had a dramatic response, their voices ceasing completely after a few sessions of the therapy. The average effect size of the therapy was 0.8. We discuss the possible psychological mechanisms for the success of Avatar therapy and the implications for the origins of persecutory voices.

How Is Individuality Expressed in Voice? An Introduction to Speech Production and Description for Speaker Classification

As well as conveying a message in words and sounds, the speech signal carries information about the speakers own anatomy, physiology, linguistic experience and mental state. These speaker characteristics are found in speech at all levels of description: from the spectral information in the sounds to the choice of words and utterances themselves. This chapter presents an introduction to speech production and to the phonetic description of speech to facilitate discussion of how speech can be a carrier for speaker characteristics as well as a carrier for messages. The chapter presents an overview of the physical structures of the human vocal tract used in speech, it introduces the standard phonetic classification system for the description of spoken gestures and it presents a catalogue of the different ways in which individuality can be expressed through speech. The chapter ends with a brief description of some applications which require access to information about speaker characteristics in speech.

Effects of noise suppression on intelligibility: dependency on signal-to-noise ratios.

The effects on speech intelligibility of three different noise reduction algorithms (spectral subtraction, minimal mean squared error spectral estimation, and subspace analysis) were evaluated in two types of noise (car and babble) over a 12 dB range of signal-to-noise ratios (SNRs). Results from these listening experiments showed that most algorithms deteriorated intelligibility scores. Modeling of the results with a logit-shaped psychometric function showed that the degradation in intelligibility scores was largely congruent with a constant shift in SNR, although some additional degradation was observed at two SNRs, suggesting a limited interaction between the effects of noise suppression and SNR.

Second-language experience and speech-in-noise recognition: Effects of talker–listener accent similarity

Previous work has shown that the intelligibility of speech in noise is degraded if the speaker and listener differ in accent, in particular when there is a disparity between native (L1) and nonnative (L2) accents. This study investigated how this talker-listener interaction is modulated by L2 experience and accent similarity. L1 Southern British English, L1 French listeners with varying L2 English experience, and French-English bilinguals were tested on the recognition of English sentences mixed in speech-shaped noise that was spoken with a range of accents (French, Korean, Northern Irish, and Southern British English). The results demonstrated clear interactions of accent and experience, with the least experienced French speakers being most accurate with French-accented English, but more experienced listeners being most accurate with L1 Southern British English accents. An acoustic similarity metric was applied to the speech productions of the talkers and the listeners, and significant correlations were obtained between accent similarity and sentence intelligibility for pairs of individuals. Overall, the results suggest that L2 experience affects talker-listener accent interactions, altering both the intelligibility of different accents and the selectivity of accent processing.

AVATAR therapy for auditory verbal hallucinations in people with psychosis: a single-blind, randomised controlled trial

Summary Background A quarter of people with psychotic conditions experience persistent auditory verbal hallucinations, despite treatment. AVATAR therapy (invented by Julian Leff in 2008) is a new approach in which people who hear voices have a dialogue with a digital representation (avatar) of their presumed persecutor, voiced by the therapist so that the avatar responds by becoming less hostile and concedes power over the course of therapy. We aimed to investigate the effect of AVATAR therapy on auditory verbal hallucinations, compared with a supportive counselling control condition. Methods We did this single-blind, randomised controlled trial at a single clinical location (South London and Maudsley NHS Trust). Participants were aged 18 to 65 years, had a clinical diagnosis of a schizophrenia spectrum (ICD10 F20–29) or affective disorder (F30–39 with psychotic symptoms), and had enduring auditory verbal hallucinations during the previous 12 months, despite continued treatment. Participants were randomly assigned (1:1) to receive AVATAR therapy or supportive counselling with randomised permuted blocks (block size randomly varying between two and six). Assessments were done at baseline, 12 weeks, and 24 weeks, by research assessors who were masked to therapy allocation. The primary outcome was reduction in auditory verbal hallucinations at 12 weeks, measured by total score on the Psychotic Symptoms Rating Scales Auditory Hallucinations (PSYRATS–AH). Analysis was by intention-to-treat with linear mixed models. The trial was prospectively registered with the ISRCTN registry, number 65314790. Findings Between Nov 1, 2013, and Jan 28, 2016, 394 people were referred to the study, of whom 369 were assessed for eligibility. Of these people, 150 were eligible and were randomly assigned to receive either AVATAR therapy (n=75) or supportive counselling (n=75). 124 (83%) met the primary outcome. The reduction in PSYRATS–AH total score at 12 weeks was significantly greater for AVATAR therapy than for supportive counselling (mean difference −3·82 [SE 1·47], 95% CI −6·70 to −0·94; p<0·0093). There was no evidence of any adverse events attributable to either therapy. Interpretation To our knowledge, this is the first powered, randomised controlled trial of AVATAR therapy. This brief, targeted therapy was more effective after 12 weeks of treatment than was supportive counselling in reducing the severity of persistent auditory verbal hallucinations, with a large effect size. Future multi-centre studies are needed to establish the effectiveness of AVATAR therapy and, if proven effective, we think it should become an option in the psychological treatment of auditory verbal hallucinations. Funding Wellcome Trust.

Using phonologically-constrained morphological analysis in continuous speech recognition

This article describes investigations into the use of phonologically-constrained morphological analysis (PCMA) in language modelling for continuous speech recognition. PCMA provides a means for modelling text as a sequence of morphemes in a way that retains compatibility with the linear concatenative model of pronunciation used in conventional decoders. Experiments were performed in English exploiting the 100-million-word British National Corpus as source material. We show that PCMA leads to smaller but more generative pronunciation lexicons, and that it does not weaken the quality of the acoustic decoding measured in terms of recognition lattices. For trigram language models, perplexity figures are poorer for PCMA over words, as might be expected given the reduction in sentence span. However recognition results show small improvements in accuracy under some conditions, particularly when morph lattices are decoded with word-trigram models. We explore the capabilities for PCMA across vocabulary size, language model training size, and post-processing strategy. The best results show a 16% relative reduction in word error rate.

ACCDIST: an accent similarity metric for accent recognition and diagnosis

ACCDIST is a metric of the similarity between speakers’ accents that is largely uninfluenced by the individual characteristics of the speakers’ voices. In this article we describe the ACCDIST approach and contrast its performance with formant and spectral-envelope similarity measures. Using a database of 14 regional accents of the British Isles, we show that the ACCDIST metric outperforms linear discriminant analysis based on either spectral-envelope or normalised formant features. Using vowel measurements from 10 male and 10 female speakers in each accent, the best spectral-envelope metric assigned the correct accent group to a held-out speaker 78.8% of the time, while the best normalised formant-frequency metric was correct 89.4% of the time. The ACCDIST metric based on spectral-envelope features, scored 92.3%. ACCDIST is also effective in clustering speakers by accent and has applications in speech technology, language learning, forensic phonetics and accent studies.

Procedures used for assessment of stuttering frequency and stuttering duration

Abstract Frequency of stuttered syllables and their durations were assessed using different procedures. The experiment examined overall syllable counts, counts of stuttered syllables and measures of stutter durations when they were made simultaneously or successively. Samples of speech with associated syllable, stuttered syllable and duration measurements of stuttering events were employed in reference transcriptions. Samples contained a minimum of 200 syllables. Ten participants assessed these samples for syllables, stuttered syllables and duration in an experiment. The responses of these participants were stored in alignment with the speech recordings for analysis. Performance was significantly more accurate (relative to transcriptions) for measures other than duration when the successive procedure was used as opposed to the simultaneous procedure. Although the successive method was more accurate, accuracy of stutter event identification was low for most participants. The procedure that allowed listeners to replay a speech sample and count the syllables, stuttered syllables and durations in three passes yielded more accurate syllable and stuttered syllable counts than procedures that required these judgments to be made in one pass.

Cue interaction in an intervocalic voiceless affricate/fricative contrast

The distinction between intervocalic voiceless affricates and fricatives is a useful one for studies of cue interaction in speech for two main reasons. First, there are at least three temporal cues that are known to interact in this contrast (the duration of the silence preceding the frication, the rise time of the frication, and the duration of the frication). Second, there have been a number of claims that auditory processes may play a crucial role both in the perception of some of the individual cues (in particular, rise time), and in accounting for the interaction of some cues (in particular, silence duration and rise time). New empirical data is presented from a multifactorial identification experiment (involving the simultaneous variation of silence duration, frication rise time and frication duration) and the extent to which various theories of affricate/fricative perception can account for the observed results is examined. [Work supported by the Joint Council Initiative in Cognitive Science/HCI (UK).]