Diphone

Abstract: A system for synthesizing a speech signal from strings of words, which are themselves strings of characters, includes a memory in which predetermined syntax tags are stored in association with entered words and phonetic transcriptions are stored in association with the syntax tags. A parser accesses the memory and groups the syntax tags of the entered words into phrases according to a first set of predetermined grammatical rules relating the syntax tags to one another. The parser also verifies the conformance of sequences of the phrases to a second set of predetermined grammatical rules relating the phrases to one another. The system retrieves the phonetic transcriptions associated with the syntax tags that were grouped into phrases conforming to the second set of rules, and also translates predetermined strings of characters into words. The system generates strings of phonetic transcriptions and prosody markers corresponding to respective strings of the words, and adds markers for rhythm and stress to the strings, which are then converted into data arrays having prosody information on a diphone-by-diphone basis. Predetermined diphone waveforms are retrieved from memory that correspond to the entered words, and these retrieved waveforms are adjusted based on the prosody information in the arrays. The adjusted diphone waveforms, which may also be adjusted for coarticulation, are then concatenated to form the speech signal. Methods in a digital computer are also disclosed.

Abstract: Synthesis Methods. Speech synthesis methods: Homage to Dennis Klatt (K.N. Stevens). Synthesis models: A discussion (E. Moulines). From data to rules (L.F.M. ten Bosch). On modelling the phonology phonetics interface for articulatory synthesis (H.J. Cedergren, G. Boulianne, D. Archambault). "Synthesis-by-rule" without segments or rewrite-rules (J. Coleman). A generation model of formant trajectories at variable speaking rates (S. Imaizumi, S. Kiritani). Spectral transitions in rule-based and diphone synthesis (D. O'Shaughnessy). On the basic scheme and algorithms in non-uniform speech synthesis (K. Takeda, K. Abe, Y. Sagisaka). Linguistic Processing. Text processing for text-to-speech synthesis (D. O'Shaughnessy). Prosodic phrasing in Swedish speech synthesis (G. Bruce, B. Granstrom, D. House). Syntactic neural networks for bi-directional text-phonetics translation (S.M. Lucas, R.I. Damper). Heuristic strategies for the higher-level analysis of unrestricted text (A.I.C. Monaghan). A framework for morphological and syntactic analysis and its application in a text-to-speech system for German (T. Russi). Novel-word pronunciation within a text-to-speech system (K.P.H. Sullivan, R.I. Damper). Prosody. Prediction of prosody: An overview (D. Hirst). A comment on the prediction of prosody (R. Collier). Automatic Learning. Syllable-based segmental duration (W.N. Campbell). Prosodic processing in a text-to-speech synthesis system using a database and learning procedures (F. Emerard, L. Mortamet, A. Cozannet). Linguistic properties in the control of segmental duration for speech synthesis (N. Kaiki, K. Takeda, Y. Sagisaka). Tree-based modelling of segmental durations (M.D. Riley). Deriving text-to-speech durations from natural speech (J.P.H. van Santen). F 0 generation with a database of natural F 0 patterns and with a neural network (C. Traber). Methodology and Models. Developing a structured lexicon for synthesis of prosody (V. Auberge). Automatic labelling of large prosodic databases: Tools, methodology and links with a text-to-speech system (G. Bailly, T. Barbe, H. Wang). A prosodic model for French text-to-speech synthesis: A psycholinguistic approach (V. Pasdeloup). Prosody and Discourse. The punctuation and perception of read and spontaneous prosody: An application to speech synthesis (I. Guaitella, S. Santi). Using discourse context to guide pitch accent decisions in synthetic speech (J. Hirschberg). System Design. Rule compilers and text-to-speech systems (G. Bailly). The MULTIVOX multilingual text-to-speech converter (G. Olaszy, G. Gordos, G. Nemeth). Concept representation for synthetic speech output system (Y. Yamashita, N. Mizutani, R. Mizoguchi). Speech Assessment. Assessment of synthetic speech (A. Fourcin). On the assessment of synthetic speech (C. Benoit, L.C.W. Pols). Segmental evaluation using the Esprit/SAM test procedures and monosyllabic words (R. Carlson, B. Granstrom, L. Nord). Segmental quality assessment by pseudo-words (A.

Abstract: A process of speech synthesis from diphones stored in a dictionary as waveforms, for text-to-speech conversion, comprises supplying a sequence of phoneme codes and respective prosodic information, and, for each phoneme, analyzing and synthesizing each phoneme, and then concatenating the synthesized phonemes. For each phoneme, two diphones are selected among the stored diphones and the presence of voicing is determined. For voiced phonemes, the respective waveforms of the two diphones constituting the phoneme are filtered by a window which is centered on a point of the selected waveform representative of the beginning of a pulse response of vocal cords to excitation thereof. The window has a width substantially equal to twice the greater of the original fundamental period and the fundamental synthesis period and has an amplitude progressively decreasing from the center of the window. The signals resulting from the filtering and obtained for each diphone are time shifted so as to be spaced apart by a time equal to the fundamental synthesis period. Synthesis is achieved by adding the displaced overlapping signals.

Abstract: A speech recognition system starts by training hidden Markov models for all triphones, diphones, and phonemes occurring in a small training vocabulary. Hidden Markov models of a target vocabulary are created by concatenating the triphone, diphone, and phoneme models, using triphone models if available, diphone HMMs when triphone models are not available, and phoneme models when neither triphone nor diphone models are available. Utterances from the target vocabulary are recognized by choosing a model with maximum probability of reproducing quantized utterance features.