Phonetic Variation Modeling and a Language Model Adaptation for Korean English Code-Switching Speech Recognition

In this paper, we propose a new method for code-switching (CS) automatic speech recognition (ASR) in Korean. First, the phonetic variations in English pronunciation spoken by Korean speakers should be considered. Thus, we tried to find a unified pronunciation model based on phonetic knowledge and deep learning. Second, we extracted the CS sentences semantically similar to the target domain and then applied the language model (LM) adaptation to solve the biased modeling toward Korean due to the imbalanced training data. In this experiment, training data were AI Hub (1033 h) in Korean and Librispeech (960 h) in English. As a result, when compared to the baseline, the proposed method improved the error reduction rate (ERR) by up to 11.6% with phonetic variant modeling and by 17.3% when semantically similar sentences were applied to the LM adaptation. If we considered only English words, the word correction rate improved up to 24.2% compared to that of the baseline. The proposed method seems to be very effective in CS speech recognition.

Macaque V1 responses to 2nd-order contrast-modulated stimuli and the possible subcortical and cortical contributions

V1 neurons as linear filters supposedly only respond to 1st-order luminance-modulated (LM) stimuli, but not 2nd-order contrast-modulated (CM) ones. To solve this difficulty, filter-rectify-filter models are proposed, in which first-stage filters respond to CM stimulus elements, and the nonlinear-rectified outputs are summed by a second-stage filter for CM stimulus representation. Correspondingly, neurophysiological evidence shows V1/A17 neurons less responsive to CM stimuli than V2/A18 neurons. Here we used two-photon calcium imaging to demonstrate substantial V1 responses to CM gratings with unimodally distributed LM/CM preferences. Moreover, LM responses are suppressed by LM and CM adaptations regardless of orientation, but CM responses are more suppressed by same-orientation LM and CM adaptations than by orthogonal ones. While LM adaptation results agree with the Hubel-Wiesel view of LGN contributions to V1 orientation responses, CM adaptation results, which include both orientation-unspecific and specific components, may suggest similar subcortical contributions plus additional refinement by recurrent intracortical interactions.

PARTIAL FILM ADAPTATIONS OF ALFRED DÖBLIN’S NOVEL “BERLIN ALEXANDERPLATZ” IN RAINER WERNER FASSBINDER’S EARLY FILMS

The film “Berlin Alexanderplatz” by Rainer Werner Fassbinder is considered to be one of the most famous literary fi lm adaptations in the fi lm history. Extensive research literature has been devoted to its study. Much less researched is the infl uence of Döblin’s novel on other works of the fi lmmaker. Th e article deals with taking over the elements of the novel “Berlin Alexanderplatz” into Fassbinder’s early fi lms. Fassbinder has borrowed constellations, plots, topoi especially those dealing with gender issues from the novel by Döblin for his fi lms “Love Is Colder Th an Death” (1969), “Gods of the Plague” (1970) and “Rio das Mortes” (1971). In the aforementioned fi lms, as in Döblin’s novel, triangular relationships are represented, the women exchange is practiced, the misogynistic protagonists exert violence against the female characters, and the woman appears as a festive link, as a copula between the men. In these fi lms, ‘Mannerbund’ in the form of criminal organizations plays an important role. Quotations and borrowings from Döblin’s novel in Fassbinder’s fi lms create a special kind of intermediality relations that diff er from fi lm adaptation in the narrow sense of the word. Following Achim von Haag the term “partial adaptation” is used here as an intensive reception of narrative structures of the literary text on the one hand and their strong modifi cation on the other.

The „Artistic Film” or the Beginnings of Adaptation

The article discusses early attempts at fi lm adaptation of literature in silent cinema, especially French. This process is connected with the establishment of the fi lm company Film d’Art and several other initiatives. A signifi cant role was fulfi lled by Société Cinématographique des Auteurs et des Hommes de Lettres (SCAGL) that functioned within the Pathé fi lm company. Its artistic director, Albert Capellani, had a connection to André Antoin, the founder of the naturalist Théâtre Libre. The infl uence of the naturalist theater can be discerned in Capellani’s adaptations of two novels by Émile Zola: – L’Assommoir and Germinal.

Language model adaptation for language and dialect identification of text

This article describes an unsupervised language model (LM) adaptation approach that can be used to enhance the performance of language identification methods. The approach is applied to a current version of the HeLI language identification method, which is now called HeLI 2.0. We describe the HeLI 2.0 method in detail. The resulting system is evaluated using the datasets from the German dialect identification and Indo-Aryan language identification shared tasks of the VarDial workshops 2017 and 2018. The new approach with LM adaptation provides considerably higher F1-scores than the basic HeLI or HeLI 2.0 methods or the other systems which participated in the shared tasks. The results indicate that unsupervised LM adaptation should be considered as an option in all language identification tasks, especially in those where encountering out-of-domain data is likely.

THE SHADOW OVER GALICIA: H.P. LOVECRAFT’S OBSESSIONS RESURFACE IN THE FILM ADAPTATION OF DAGON (2001)

With the cinematic adaptation of H.P. Lovecraft’s tale The Shadow Over Innsmouth (1931) under the title of Dagon (Stuart Gordon, 2001), many ideological issues which have haunted scholarly appreciation of his literary work have been brought out under a new light. The setting of the lm in the Galician coast of Spain, from which many immigrants crossed the ocean towards America, provides a further reading to the hardly concealed xenophobia underlying this and many other tales. Hybridity, linguistic plurality, sexual taboo and other controversial features are updated far more explicitly on screen, proving that Lovecraft’s extremist perspective is not a thing of the past.Keywords: North-American Literature, lm adaptation, H.P. Lovecraft, Dagon, The Shadow Over Innsmouth, ethnic stereotypes, popular culture, immigration.Título en español: La sombra sobre Galicia: las obsesiones de H.P. Lovecraft resurgen en la adaptación cinematográfica de Dagon (2001)Resumen: Al adaptar al cine el relato de H.P. Lovecraft La sombra sobre Innsmouth (1931) con el título de Dagon (2001), Stuart Gordon genera nuevas posibilidades de interpretación del texto original al transferir la acción a España, concretamente al pueblo ficticio de Imboca (obvia derivación del original inglés) en la costa gallega. Una decisión puramente comercial activa diversas cargas semánticas subyacentes tradicionalmente adscritas a la obra del autor y vinculadas a su firme rechazo a la emigración hacia Estados Unidos. Estereotipos nacionales, hibridación, xenofobia y otros aspectos controvertidos vinculados a la localización de la película con rman que el extremismo político de Lovecraft sobrevive al cambio de siglo.Palabras clave: Literatura norteamericana, adaptación cinematográ ca, H.P. Lovecraft, Dagon, La sombra sobre Innsmouth, estereotipos nacionales, cultura popular, emigración.

Improvement of Lecture Speech Recognition by Using Unsupervised Adaptation

The aim of this work is to improve the recognition performance of spontaneous speech. In order to achieve the purpose, the authors of this chapter propose new approaches of unsupervised adaptation for spontaneous speech and evaluate the methods by using diagonal-covariance and full-covariance hidden Markov models. In the adaptation procedure, both methods of language model (LM) adaptation and acoustic model (AM) adaptation are used iteratively. Several combination methods are tested to find the optimal approach. In the LM adaptation, a word trigram model and a part-of-speech (POS) trigram model are combined to build a more task-specific LM. In addition, the authors propose an unsupervised speaker adaptation technique based on adaptation data weighting. The weighting is performed depending on POS class. In Japan, a large-scale spontaneous speech database “Corpus of Spontaneous Japanese (CSJ)” has been used as the common evaluation database for spontaneous speech and the authors used it for their recognition experiments. From the results, the proposed methods demonstrated a significant advantage in that task.