Syllable language models for Mandarin speech recognition: exploiting character language models.

Mandarin Chinese is based on characters which are syllabic in nature and morphological in meaning. All spoken languages have syllabiotactic rules which govern the construction of syllables and their allowed sequences. These constraints are not as restrictive as those learned from word sequences, but they can provide additional useful linguistic information. Hence, it is possible to improve speech recognition performance by appropriately combining these two types of constraints. For the Chinese language considered in this paper, character level language models (LMs) can be used as a first level approximation to allowed syllable sequences. To test this idea, word and character level n-gram LMs were trained on 2.8 billion words (equivalent to 4.3 billion characters) of texts from a wide collection of text sources. Both hypothesis and model based combination techniques were investigated to combine word and character level LMs. Significant character error rate reductions up to 7.3% relative were obtained on a state-of-the-art Mandarin Chinese broadcast audio recognition task using an adapted history dependent multi-level LM that performs a log-linearly combination of character and word level LMs. This supports the hypothesis that character or syllable sequence models are useful for improving Mandarin speech recognition performance.

Ontogeny of IgM-producing cells in the mandarin fish Siniperca chuatsi identified by in situ hybridisation

The ontogeny of IgM-producing cells was studied in juvenile mandarin fish Simperca chuatsi, an important fish in China's aquaculture sector. The IgM-producing cells were localised through in situ hybridisation with a probe complementary to the Ig mu-chain in lymphoid-related tissues, including head kidney, spleen, thymus, intestine and gills. In head kidney, transcripts of Ig mu were first detected at 20 days post-hatching (dph) with a few positive signals. and the number of IgM-producing cells increased obviously from 39 dph onwards. At 136 dph, a large amount of positive cells were observed in the entire organ with clusters of these cells located around the blood vessels. In spleen, IgM-producing cells were found from 26 dph onwards, followed by an increase until 67 dph: clusters of positive cells were also detected around blood vessels at 102 dph. In thymus, IgM-producing cells were first observed at 39 dph; thereafter, no obvious increase was detected until 78 dph. The positive cells in thymus were distributed mainly in the outer zone of thymus. A few IgM-producing cells were still observed in thymus of 1-year-old mandarin fish. IgM-producing cells were not detected in the intestine until 87 dph, with several discrete positively stained cells distributed in the lamina propria. IgM-producing cells, scattered mainly in primary gill filaments around blood vessels, were detected in gills from 90 dph. As in other teleosts, these results indicated that the head kidney appears to be the primary organ for IgM production in mandarin fish, and IgM-producing cells exist in all organs examined in the present study, implying their lymphoid role in fish. In addition, it is suggested that vaccination after 20 dph may be much more effective in mandarin fish. (C) 2009 Elsevier B.V. All rights reserved.