We present an attention-based sequence-to-sequence neural network which can directly translate speech from one language into speech in another language, without relying on an intermediate text representation. The network is trained end-to-end, learning to map speech spectrograms into target spectrograms in another language, corresponding to the translated content (in a different canonical voice). We further demonstrate the ability to synthesize translated speech using the voice of the source speaker. We conduct experiments on two Spanish-to-English speech translation datasets, and find that the proposed model slightly underperforms a baseline cascade of a direct speech-to-text translation model and a text-to-speech synthesis model, demonstrating the feasibility of the approach on this very challenging task.
We present a classifier to predict contextual polarity of subjective phrases in a sentence. Our approach features lexical scoring derived from the Dictionary of Affect in Language (DAL) and extended through WordNet, allowing us to automatically score the vast majority of words in our input avoiding the need for manual labeling. We augment lexical scoring with n-gram analysis to capture the effect of context. We combine DAL scores with syntactic constituents and then extract ngrams of constituents from all sentences. We also use the polarity of all syntactic constituents within the sentence as features. Our results show significant improvement over a majority class baseline as well as a more difficult baseline consisting of lexical n-grams.
The Arabic language is a collection of multiple variants, among which Modern Standard Arabic (MSA) has a special status as the formal written standard language of the media, culture and education across the Arab world. The other variants are informal spoken dialects that are the media of communication for daily life. Arabic dialects differ substantially from MSA and each other in terms of phonology, morphology, lexical choice and syntax. In this paper, we describe a system that automatically identifies the Arabic dialect (Gulf, Iraqi, Levantine, Egyptian and MSA) of a speaker given a sample of his/her speech. The phonotactic approach we use proves to be effective in identifying these dialects with considerable overall accuracy -81.60% using 30s test utterances.
We describe Parrotron, an end-to-end-trained speech-to-speech conversion model that maps an input spectrogram directly to another spectrogram, without utilizing any intermediate discrete representation. The network is composed of an encoder, spectrogram and phoneme decoders, followed by a vocoder to synthesize a time-domain waveform. We demonstrate that this model can be trained to normalize speech from any speaker regardless of accent, prosody, and background noise, into the voice of a single canonical target speaker with a fixed accent and consistent articulation and prosody. We further show that this normalization model can be adapted to normalize highly atypical speech from a deaf speaker, resulting in significant improvements in intelligibility and naturalness, measured via a speech recognizer and listening tests. Finally, demonstrating the utility of this model on other speech tasks, we show that the same model architecture can be trained to perform a speech separation task.
In this paper, we show that linguistically motivated pronunciation rules can improve phone and word recognition results for Modern Standard Arabic (MSA). Using these rules and the MADA morphological analysis and disambiguation tool, multiple pronunciations per word are automatically generated to build two pronunciation dictionaries; one for training and another for decoding. We demonstrate that the use of these rules can significantly improve both MSA phone recognition and MSA word recognition accuracies over a baseline system using pronunciation rules typically employed in previous work on MSA Automatic Speech Recognition (ASR). We obtain a significant improvement in absolute accuracy in phone recognition of 3.77%-7.29% and a significant improvement of 4.1% in absolute accuracy in ASR.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.