Search

Impulse magnetron discharges with millisecond-scale pulses—extended duration modes of technological high-power impulse magnetron sputtering (HiPIMS)—have promising applications in pulsed plasma facilities of different kinds due to high ionization degree (up to 90%), suitable duration, and scalable hardware design. Depending on operating conditions, at the same power level, two distinct diffuse regimes can be distinguished: the one with intense target sputtering— long HiPIMS (L-HiPIMS),—and the non-sputtering low-voltage one (non-sputtering magnetron discharge). The majority of existing studies of these discharge forms were made for argon working gas. For a number of prospective high-power pulsed plasma applications (e. g. pulsed plasma accelerators and thrusters), however, the option of using light gases is preferable. Here, the operation of a millisecond-scale impulse magnetron discharge (L-HiPIMS) in helium has been examined. The plasma parameters (electron density, electron temperature) were measured in a time-resolved fashion with a Langmuir probe. The electrical measurements were accompanied with optical emission spectroscopy. The use of the long pulsed modes enables achieving high plasma density and accelerating the ion flux with a peak energy of ~ 10 eV.

We compared two modes of magnetron sputter deposition of MoS2 on substrates made of steel AISI 316L and bronze CuAl9NiFe4Mn1 with different initial roughness Ra 0.05–2.32 µm. The deposition was carried out at a bias voltage of −20 and +100 V, and the deposition rate of these modes differed by 30%. Measurements of the friction coefficient and lifetime tests were made in accordance with ASTM G133. Measurements of the friction coefficient and lifetime tests in vacuum at load 7H, temperature of samples 250 °C, and pressure in the chamber

Background Augmentative and alternative communication (AAC) can assist persons with complex communication needs to communicate competently with a variety of communication partners in a variety of contexts. However, AAC systems and intervention often do not take multilingual aspects into consideration. Objective This small-scale exploratory study had three aims, namely: (1) to describe the self-reported language skills of multilingual South African adults using AAC, (2) to describe the languages and communication modalities they used in interaction and (3) to obtain their views regarding access to various languages. Methods Twenty-seven adults using AAC were recruited via an empowerment programme, as well as an email list for persons interested in AAC, and provided responses to a questionnaire. To compensate for access and written language challenges, the questionnaire was administered with help and/or as a face-to-face interview where needed. Responses were analysed using mostly descriptive statistics. Results Participants generally could not express themselves in all the languages they understood and were regularly exposed to. Speech-generating devices specifically gave access almost exclusively to English. Participants expressed a desire to increase their expressive language repertoire, and mentioned both limitations of communication technology as well as their own literacy skills as barriers to overcome in this regard. Conclusion In order for multilingual South African adults using AAC to express themselves in multiple languages, appropriate AAC systems and interventions as well as literacy learning opportunities need to be developed and provided. Keywords adult, augmentative and alternative communication; multilingualism; complex communication needs, language and communication skills; self-report, views.

This work explores the effect of input text representation on the performance of a Long Short-Term Memory Recurrent Neural Network Language Model (LSTM RNNLM). Due to a problem with vanishing gradients during LSTM RNNLM training, they cannot capture long-range context information. Long-range context information often encapsulates details about the text application domain. Word embedding vectors capture similarity and semantic context information in their structure. Additional context can be captured by socio-situational setting information, topics, named entities and parts-of-speech tags. This work uses a character LSTM RNNLM as a control in experiments to determine the effect different types of input text representation have on the perplexity of an LSTM RNNLM in terms of a percentage increase or decrease in the value. Adding socio- situational information to a character LSTM RNNLM results in a 0.1 % reduction in perplexity in comparison to that of the control model. When the character embeddings are swapped with word2vec embeddings a reduction in perplexity of 2.77 % is obtained. Adding context information such as socio-situational information to the word embedded model should also result in a perplexity reduction. However, this is not the case, as the addition of socio-situational information to a word embedded model results in a 5.79 % perplexity increase in comparison to the word2vec only model. This trend of an increase in perplexity is observed in further experiments where other types of context information are added to a word embedded model. The largest increase in perplexity is obtained when word embeddings and topics are applied giving a perplexity increase of 7.55 %. This increase in perplexity is due to the addition of more data (context information) to the input text. More data means more words (unique or otherwise) that are being concatenated together as a representation of the input. This results in a larger and sparser input that not only takes longer to train but has less useful information captured on average resulting in models with a higher perplexity. A better method of text representation that will reduce the size of the input while still capturing the necessary semantic information implicit in word embeddings will be adding the vectors together instead of concatenating them.

We report on two sets of perceptual evaluations of our South African text-to-speech voices by language practitioners. In the first evaluation, we measure baseline quality in terms of how understandable and human-like the voices sound. We also determine baseline usability by asking a series of questions related to accessibility and mainstream application settings. In the second evaluation, we employ the same criteria to compare pronunciation improvements against the baseline. The results indicate success in many areas, but also illuminate room for improvement in others, especially in the cases of the African languages.

Text normalisation in text-to-speech synthesis comprises the segmentation and classification of the incoming text and the subsequent expansion of non-standard words into their standard word, spoken forms. We present a rule-based implementation for the 11 official South African languages that uses native number expansion. We discuss the architecture and performance of the implementation based on examples of cardinal and ordinal numbers, money, dates and times. Although the implementation scores well, it is currently limited to handling non-standard words in isolation. Future work will need to address sentence context in order to normalise the African languages correctly.

In South Africa, many persons in need of augmentative and alternative communication (AAC) come from multilingual backgrounds. There is as yet a limited evidence base (locally and internationally) for the provision of AAC services to multilingual populations. The perspectives of service providers can assist in gaining an understanding of current practices and the factors that influence these.The study aimed to obtain the perspectives of AAC service providers about practices in providing AAC systems and AAC intervention to clients from multilingual backgrounds.Fifteen AAC service providers were purposefully chosen to participate in one of three focus groups - two face-to-face and one online focus group. Data from the face-to-face focus groups was transcribed verbatim. Thematic analysis was used to identify themes and subthemes in the data.Four overarching themes were identified, namely (a) current practices, (b) factors influencing current practices, (c) service provides' orientation towards different language options in AAC intervention, and (d) needs and desired developments regarding AAC technology. This paper reports on the first three themes. Service providers reported their practices to range from a focus on L1 exclusively, L2 exclusively, to a multilingual (sequential or simultaneous) approach. The South African language context, family language preferences and choices, service provider skill and knowledge, as well as AAC technology were identified as factors influencing their practices. Although many viewed access to multiple languages through AAC as positive, they also expressed concerns and reservations about providing multilingual AAC services.Although service providers in general saw the need to give clients from multilingual backgrounds access to multiple languages using AAC, this did not always translate into multilingual AAC practices. Both extrinsic factors (e.g. the lack of appropriate AAC devices, software and apps giving access to non-English languages) and intrinsic factors (service providers' language competency and their beliefs about the cognitive demands of multilingual AAC systems) influenced their practices and choices. Appropriate AAC service delivery to multilingual populations in South Africa would require not only appropriate AAC technology developments, but also research evidence to establish the efficacy of multilingual AAC interventions for clients with a variety of characteristics.