Documenting sound change with smartphone apps

Crowdsourcing linguistic phenomena with smartphone applications is relatively new. Apps have been used to train acoustic models for automatic speech recognition (de Vries et al. 2014) and to archive endangered languages (Iwaidja Inyaman Team 2012). Leemann and Kolly (2013) developed a free app for iOS—Dialäkt Äpp (DÄ) (>78k downloads)—to document language change in Swiss German. Here, we present results of sound change based on DÄ data. DÄ predicts the users’ dialects: for 16 variables, users select their dialectal variant. DÄ then tells users which dialect they speak. Underlying this prediction are maps from the Linguistic Atlas of German-speaking Switzerland (SDS, 1962-2003), which documents the linguistic situation around 1950. If predicted wrongly, users indicate their actual dialect. With this information, the 16 variables can be assessed for language change. Results revealed robustness of phonetic variables; lexical and morphological variables were more prone to change. Phonetic variables like to lift (variants: /lupfə, lʏpfə, lipfə/) revealed SDS agreement scores of nearly 85%, i.e., little sound change. Not all phonetic variables are equally robust: ladle (variants: /xælə, xællə, xæuə, xæɫə, xæɫɫə/) exhibited significant sound change. We will illustrate the results using maps that show details of the sound changes at hand.

A Food Recognition System for Diabetic Patients based on an Optimized Bag of Features Model

Computer vision-based food recognition could be used to estimate a meal's carbohydrate content for diabetic patients. This study proposes a methodology for automatic food recognition, based on the Bag of Features (BoF) model. An extensive technical investigation was conducted for the identification and optimization of the best performing components involved in the BoF architecture, as well as the estimation of the corresponding parameters. For the design and evaluation of the prototype system, a visual dataset with nearly 5,000 food images was created and organized into 11 classes. The optimized system computes dense local features, using the scale-invariant feature transform on the HSV color space, builds a visual dictionary of 10,000 visual words by using the hierarchical k-means clustering and finally classifies the food images with a linear support vector machine classifier. The system achieved classification accuracy of the order of 78%, thus proving the feasibility of the proposed approach in a very challenging image dataset.

Speech Intelligibility in Noise With a Pinna Effect Imitating Cochlear Implant Processor.

OBJECTIVE To evaluate the speech intelligibility in noise with a new cochlear implant (CI) processor that uses a pinna effect imitating directional microphone system. STUDY DESIGN Prospective experimental study. SETTING Tertiary referral center. PATIENTS Ten experienced, unilateral CI recipients with bilateral severe-to-profound hearing loss. INTERVENTION All participants performed speech in noise tests with the Opus 2 processor (omnidirectional microphone mode only) and the newer Sonnet processor (omnidirectional and directional microphone mode). MAIN OUTCOME MEASURE The speech reception threshold (SRT) in noise was measured in four spatial settings. The test sentences were always presented from the front. The noise was arriving either from the front (S0N0), the ipsilateral side of the CI (S0NIL), the contralateral side of the CI (S0NCL), or the back (S0N180). RESULTS The directional mode improved the SRTs by 3.6 dB (p < 0.01), 2.2 dB (p < 0.01), and 1.3 dB (p < 0.05) in the S0N180, S0NIL, and S0NCL situations, when compared with the Sonnet in the omnidirectional mode. There was no statistically significant difference in the S0N0 situation. No differences between the Opus 2 and the Sonnet in the omnidirectional mode were observed. CONCLUSION Speech intelligibility with the Sonnet system was statistically different to speech recognition with the Opus 2 system suggesting that CI users might profit from the pinna effect imitating directionality mode in noisy environments.

Audiological results with Baha in conductive and mixed hearing loss

The level of improvement in the audiological results of Baha(®) users mainly depends on the patient's preoperative hearing thresholds and the type of Baha sound processor used. This investigation shows correlations between the preoperative hearing threshold and postoperative aided thresholds and audiological results in speech understanding in quiet of 84 Baha users with unilateral conductive hearing loss, bilateral conductive hearing loss and bilateral mixed hearing loss. Secondly, speech understanding in noise of 26 Baha users with different Baha sound processors (Compact, Divino, and BP100) is investigated. Linear regression between aided sound field thresholds and bone conduction (BC) thresholds of the better ear shows highest correlation coefficients and the steepest slope. Differences between better BC thresholds and aided sound field thresholds are smallest for mid-frequencies (1 and 2 kHz) and become larger at 0.5 and 4 kHz. For Baha users, the gain in speech recognition in quiet can be expected to lie in the order of magnitude of the gain in their hearing threshold. Compared to its predecessor sound processors Baha(®) Compact and Baha(®) Divino, Baha(®) BP100 improves speech understanding in noise significantly by +0.9 to +4.6 dB signal-to-noise ratio, depending on the setting and the use of directional microphone. For Baha users with unilateral and bilateral conductive hearing loss and bilateral mixed hearing loss, audiological results in aided sound field thresholds can be estimated with the better BC hearing threshold. The benefit in speech understanding in quiet can be expected to be similar to the gain in their sound field hearing threshold. The most recent technology of Baha sound processor improves speech understanding in noise by an order of magnitude that is well perceived by users and which can be very useful in everyday life.

Screening for Language Disorders in Stroke: German Validation of the Language Screening Test (LAST)

BACKGROUND Screening of aphasia in acute stroke is crucial for directing patients to early language therapy. The Language Screening Test (LAST), originally developed in French, is a validated language screening test that allows detection of a language deficit within a few minutes. The aim of the present study was to develop and validate two parallel German versions of the LAST. METHODS The LAST includes subtests for naming, repetition, automatic speech, and comprehension. For the translation into German, task constructs and psycholinguistic criteria for item selection were identical to the French LAST. A cohort of 101 stroke patients were tested, all of whom were native German speakers. Validation of the LAST was based on (1) analysis of equivalence of the German versions, which was established by administering both versions successively in a subset of patients, (2) internal validity by means of internal consistency analysis, and (3) external validity by comparison with the short version of the Token Test in another subset of patients. RESULTS The two German versions were equivalent as demonstrated by a high intraclass correlation coefficient of 0.91. Furthermore, an acceptable internal structure of the LAST was found (Cronbach's α = 0.74). A highly significant correlation (r = 0.74, p < 0.0001) between the LAST and the short version of the Token Test indicated good external validity of the scale. CONCLUSION The German version of the LAST, available in two parallel versions, is a new and valid language screening test in stroke.

Segmentation and Recognition of Multi-Food Meal Images for Carbohydrate Counting

In this paper, we propose novel methodologies for the automatic segmentation and recognition of multi-food images. The proposed methods implement the first modules of a carbohydrate counting and insulin advisory system for type 1 diabetic patients. Initially the plate is segmented using pyramidal mean-shift filtering and a region growing algorithm. Then each of the resulted segments is described by both color and texture features and classified by a support vector machine into one of six different major food classes. Finally, a modified version of the Huang and Dom evaluation index was proposed, addressing the particular needs of the food segmentation problem. The experimental results prove the effectiveness of the proposed method achieving a segmentation accuracy of 88.5% and recognition rate equal to 87%

Automatic identification of discourse markers in dialogues: An in-depth study of like and well

The lexical items like and well can serve as discourse markers (DMs), but can also play numerous other roles, such as verb or adverb. Identifying the occurrences that function as DMs is an important step for language understanding by computers. In this study, automatic classifiers using lexical, prosodic/positional and sociolinguistic features are trained over transcribed dialogues, manually annotated with DM information. The resulting classifiers improve state-of-the-art performance of DM identification, at about 90% recall and 79% precision for like (84.5% accuracy, κ = 0.69), and 99% recall and 98% precision for well (97.5% accuracy, κ = 0.88). Automatic feature analysis shows that lexical collocations are the most reliable indicators, followed by prosodic/positional features, while sociolinguistic features are marginally useful for the identification of DM like and not useful for well. The differentiated processing of each type of DM improves classification accuracy, suggesting that these types should be treated individually.

Towards Automatic Identification of Discourse Markers in Dialogs: the Case of 'Like'

This article discusses the detection of discourse markers (DM) in dialog transcriptions, by human annotators and by automated means. After a theoretical discussion of the definition of DMs and their relevance to natural language processing, we focus on the role of like as a DM. Results from experiments with human annotators show that detection of DMs is a difficult but reliable task, which requires prosodic information from soundtracks. Then, several types of features are defined for automatic disambiguation of like: collocations, part-of-speech tags and duration-based features. Decision-tree learning shows that for like, nearly 70% precision can be reached, with near 100% recall, mainly using collocation filters. Similar results hold for well, with about 91% precision at 100% recall.