Diphone Synthesis Research Articles

Speech Synthesis using Vocabulary of Interphoneme Transitions

Objectives: This paper presents a modified method of diphone synthesis of Kazakh language. The aim of this work is to create a high-speed synthesizer with acceptable quality. Methods/Statistical Analysis: Diphones represent transitions between sounds (interphoneme transitions). Therefore, to increase the quality of standard method, a technique was used for trimming the edges according to signal quasi-periods and automatic splicing normalization according to the amplitude at the beginning and the ends, which gave a significant increase in quality while maintaining the advantages of this method of synthesis. Application/Improvements: Finally, the described method is implemented as software for speech synthesis of the Kazakh and Russian languages. Keywords: Diphone, Quasi-Period, Phoneme, Synthesis, Speech

English

In this paper, we describe the design and develop a database of Indonesian diphone synthesis using speech segment of recorded voice to be converted from text to speech and save it as audio file like WAV or MP3. In designing and develop a database of Indonesian diphone there are several steps to follow; First, developed Diphone database includes: create a list of sample of words consisting of diphones organized by prioritizing looking diphone located in the middle of a word if not at the beginning or end; recording the samples of words by segmentation. ;create diphones made with a tool Diphone Studio 1.3. Second, develop system using Microsoft Visual Delphi 6.0, includes: the conversion system from the input of numbers, acronyms, words, and sentences into representations diphone. There are two kinds of conversion (process) alleged in analyzing the Indonesian text-to-speech system. One is to convert the text to be sounded to phonem and two, to convert the phonem to speech. Method used in this research is called Diphone Concatenative synthesis, in which recorded sound segments are collected. Every segment consists of a diphone (2 phonems). This synthesizer may produce voice with high level of naturalness. The Indonesian Text to Speech system can differentiate special phonemes like in ‘Beda’ and ‘Bedak’ but sample of other spesific words is necessary to put into the system. This Indonesia TTS system can handle texts with abbreviation, there is the facility to add such words. Keywords : diphone; text to speech; concatenative synthesis.

Subjective responses to synthesised speech with lexical emotional content: the effect of the naturalness of the synthetic voice

This study aimed to investigate how the degree of naturalness and lexical emotional content of synthesised speech affects the subjective ratings of emotional experiences and how the naturalness of the voice affects the ratings of voice quality. Twenty-four participants listened to a set of affective words produced by three different speech synthesis techniques: formant synthesis, diphone synthesis and unit selection synthesis. The participants’ task was to rate their experiences evoked by the speech samples using three emotion-related bipolar scales for valence, arousal and approachability. The pleasantness, naturalness and clarity of the voices were also rated. The results showed that the affective words produced by the synthesisers evoked congruent emotion-related ratings in the participants. The ratings of the experienced valence and approachability were statistically significantly stronger when the affective words were produced by the more humanlike voices as compared to the more machinelike voice. The more humanlike voices were also rated as statistically significantly more natural, pleasant and clear than the less humanlike voice. Thus, our findings suggest that even machinelike voices can be used to communicate affective messages but that increasing the level of naturalness enhances positive feelings about synthetic voices and strengthens emotional communication between computers and humans.

On the intelligibility of fast synthesized speech for people who are blind: A cross-system comparison.

People who are blind increasingly use synthesized speech as a primary output modality when interacting with computers, mobile devices, and web-based services. They typically prefer to listen to the synthesized speech at speeds multiple times real time, and consequently develop strong preferences for particular text-to-speech (TTS) engines and voices. These usage-based preferences lead to potentially incorrect assumptions about which TTS approaches are “best” for fast speech. We report on a cross-system comparison of the intelligibility of fast synthesized speech for users who are blind from birth. We used male and female voices from multiple TTS engines representing the main approaches to TTS: diphone synthesis, unit selection synthesis, HMM-based synthesis, and formant synthesis. We recruited participants from organizations that work with the blind. Each participant listened to and transcribed semantically underspecified sentences from a single TTS engine and voice, spoken at speeds ranging from 300 to 550 words/min. We used TTS engine, algorithm for speeding up synthesized speech, and speaker sex as independent variables, and transcription accuracy as the primary dependent variable. We report differences by TTS approach, speed up method, TTS engine type, and speaker sex.

Expressive speech synthesis using multilevel unit selection

Generating natural sounding and meaningful prosody is a central challenge in text-to-speech synthesis, especially when generating expressive speech. Recently, we proposed a multilevel unit sequence synthesis approach, based on the general superpositional model of intonation, which describes a pitch contour as the sum of component curves that are each associated with different phonological levels, specifically the phoneme, foot, and phrase. During synthesis, segmental perturbation curves, accent curves, and phrase curves are extracted from the acoustic signal and are combined into target pitch curves; these target curves are then imposed on the acoustic unit sequences using standard pitch modification methods. This approach represents an attempt to combine the strengths of the two dominant approaches to speech synthesis: unit selection synthesis, which preserves all details of natural speech but struggles with coverage of the very large combinatorial space of phoneme sequences and prosodic contexts, and diphone synthesis, which addresses coverage by generating rule-based synthetic target prosody and imposing it on acoustic units using signal modification methods. This approach minimizes prosodic modification artifacts, optimizes the naturalness of the target pitch contour by using quasi-natural contours, yet avoids the combinatorial explosion of standard unit selection synthesis.

Database size and naturalness in concatenative speech synthesis

Unit concatenation TTS systems seek to maximize perceived naturalness by minimizing the amount of signal processing applied to the recorded speech on which they are based. To generate distinct suprasegmentals for a given segmental sequence (e.g., to convey variation in focus or emotion), it is necessary to record and store multiple instances of the same segments that vary in fundamental frequency and voice quality. At the expense of naturalness, concatenative systems can store a minimal segmental inventory and synthesize suprasegmental factors by manipulating f0 and voice quality via signal processing. Classic diphone synthesis (where only a single instance of each diphone sequence is stored) represents the limiting case of this strategy. The present study explores aspects of the trade‐off between perceived naturalness and segmental inventory size using the ModelTalker TTS system. Twenty‐five speakers each recorded about 1650 utterances. From these, databases were constructed that limited the maximum number of alternate diphones usable for synthesis in five conditions to 1, 5, 10, 20, and 40. Sentences were synthesized from these databases using either full or no f0 control. Results of listening tests wherein subjects rate the naturalness of each sentence will be presented. [Work supported by NIDCD.]

Reducing audible spectral discontinuities

A common problem in diphone synthesis is discussed, viz., the occurrence of audible discontinuities at diphone boundaries. Informal observations show that spectral mismatch is the most likely the clause of this phenomenon. We first set out to find an objective spectral measure for discontinuity. To this end, several spectral distance measures are related to the results of a listening experiment. Then, we studied the feasibility of extending the diphone database with context-sensitive diphones to reduce the occurrence of audible discontinuities. The number of additional diphones is limited by clustering consonant contexts that have a similar effect on the surrounding vowels on the basis of the best performing distance measure. A listening experiment has shown that the addition of these context-sensitive diphones significantly reduces the amount of audible discontinuities.

An On-Line Computer-Controlled Diphone Synthesis System

Speech synthesis by diphone segmentation consists of the stylization and simulation of the perceptually meaninfgul, dynamic portions of the acoustic continuum of speech. Diphone segment assembly is a technique for synthesizing a potentially unlimited variety of continuous utterances under computer control. Investigations of this synthesis technique have progressed until the previous methods of off-line computer processing, and punch card input and storage are no longer capable of dealing with the bulk and variety of data that must be handled in order to examine adequately the synthesis of General American English speech. The new diphone synthesis system to be described consists of several components that are compatible with the IBM System/360. The present combination consists of an IBM 360/40 System, an IBM 2250 Graphic Display Terminal, and a digitally controlled terminal analog speech synthesizer.

Some Effects of List Familiarity and Synthetic Segment Familiarity on the Intelligibility of PB Words Produced by Diphone Synthesis

In a previous paper [S. E. Estes, H. R. Kerby, H. D. Maxey, and R. M. Walker, “Speech Synthesis from Stored Data,” J. Acoust. Soc. Am. 34, 2003(A) (1962)], a method for unlimited vocabulary synthesis was presented and discussed. The method consists of computer assembly of digital data representing diphone-length segments, together with auxiliary data specifying intonation and timing, to produce synthetic speech. In order to evaluate the adequacy of this method of segmentation, intelligilibity tests have been conducted using PB words in a standard carrier phrase. The results of these studies have indicated that intelligibility scores on synthetically produced PB words are affected by at least two major factors: familiarity with the words under test and familiarity with the diphone components of the words. A study has been designed to test the effects of these two factors on the intelligibility scores of normal adult listeners in sound-field conditions. The stimuli used are selected C.I.D. W-22 lists. These results are presented and discussed.