TTS Engine Research Articles

Raspberry Pi Based Reader for Blind

The Braille machines are expensive and as a result, are not accessible to many. In particular, there is a need for a portable text reader that is affordable and readily available to the blind community. The designed system is a portable, low-cost, reading device made for blind people. The designed system also overcomes the limitation of a conventional Braille machine by making it affordable for the common masses. This paper proposes a smart reader for visually challenged people using raspberry pi. This paper addresses the integration of a complete Text Read-out system designed for the visually challenged. The designed system consists of a webcam interfaced with raspberry pi which accepts a page of printed text. This system uses OCR technology to convert images into text and reads out the text by using Text-to-Speech conversion. This system supports audio output via Speakers as well as headphones. The OCR (Optical Character Recognition) package is installed in raspberry pi which scans it into a digital document which is then subjected to skew correction, and segmentation, before feature extraction to perform classification. Once classified, the text is read out by a text-to-speech conversion unit (TTS engine) installed in raspberry pi. The output is fed to an audio amplifier before it is read out.

Smart Assisting for Visually Impaired People

Project addresses the integration of a complete Text Read-out system designed for the visually challenged. The system consists of a web cam interfaced with Raspberry Pi which accepts a page of printed text. The OCR package installed in Raspberry Pi, scans it into a digital document which is then subjected to image processing like skew correction, segmentation to classify the text that is captured. After the classification, the text is readout by a text to speech conversion unit (TTS engine) installed in Raspberry Pi. The output is fed to an audio amplifier then it is read out. The simulation is just an initiation of the image to text conversion and text to speech conversion done by the OCR software installed in Raspberry Pi. The system finds interesting applications in libraries, auditoriums, offices where instructions and notices are to be read and also in the assisted paper documents. By using ultrasonic sensor, we will measure the distance between the blind people and obstacle then the distance will be played through audio output..

TTS-driven Embodied Conversation Avatar for UMB-SmartTV

When human-TV interaction is performed by remote controller and mobile devices only, the interactions tend to be mechanical, dreary and uninformative. To achieve more advanced interaction, and more human-human like, we introduce the virtual agent technology as a feedback interface. Verbal and co-verbal gestures are linked through complex mental processes, and although they represent different sides of the same mental process, the formulations of both are quite different. Namely, verbal information is bound by rules and grammar, whereas gestures are influenced by emotions, personality etc. In this paper a TTS-driven behavior generation system is proposed for more advanced interface used for smart IPTV platforms. The system is implemented as a distributive non-IPTV service and integrated into UMB-SmartTV in a service-oriented fashion. The behavior generation system fuses speech and gesture production models by using FSMs and HRG structures. Features for selecting the shape and alignment of co-verbal movement are based on linguistic features (that can be extracted from arbitrary input text), and prosodic features (as predicted within several processing steps in the TTS engine). At the end, the generated speech and co-verbal behavior are animated by an embodied conversational agent (ECA) engine and represented to the user within the UMB-SmarTV user interface.

Kannada Text to Speech Conversion System

The following paper describes the design of a system which does text to speech generation for one of the regional language’s Kannada. The printed document of Kannada text is given as input to the system, the system then converts the document to an image format. Pre-processing is done to stabilize the intensity of the images and clear the artifacts. This process boosts the precision and interpretability of an image. Optical Character Recognition (OCR) is used to unsheathe the segmented characters from a particular image and are matched with the characters that have been stored in the dataset. Once the matched characters are extracted it is stored in a suitable format and then the TTS engine is deployed to convert the saved Kannada characters to a speech format. The obtained speech output corresponds to the characters which are collected after processing the input text.

The First Vietnamese FOSD-Tacotron-2-based Text-to-Speech Model Dataset.

Recent trends in voicebot application development have enabled utilization of both speech-to-text and text-to-speech (TTS) generation techniques. In order to generate a voice response to a given speech, one needs to use a TTS engine. The recently developed TTS engines are shifting towards end-to-end approaches utilizing models such as Tacotron, Tacotron-2, WaveNet, and WaveGlow. The reason is that it enables a TTS service provider to focus on developing training and validating datasets comprising of labelled texts and recorded speeches instead of designing an entirely new model that outperforms the others which is time-consuming and costly. In this context, this work introduces the first Vietnamese FPT Open Speech Data (FOSD)-Tacotron-2-based TTS model dataset. This dataset comprises of a configuration file in *.json format; training and validating text input files (in *.csv format); a 225,000-step checkpoint of the trained model; and several sample generated audios. The published dataset is extremely worth for serving as a model for benchmarking with other newly developed TTS models / engines. In addition, it opens an entirely new TTS research optimization problem to be addressed: How to effectively generate speech from text given: a black box TTS (trained) model and its training and validation input texts.

TRANSFORMING INTERACTIONS: MOUSE-BASED TO VOICE-BASED INTERFACES

Recent technological advancements in natural language processing and voice-based technology have led to transformations in the way humans, and machines interact. From the conventional mouse-based interactions, the demand has switched over to more advanced voice-based interactions. The naturalness of speech as a communication medium coupled with technological advancements has influenced this demand. This paper presents an application of computers and technology to develop a voice-based interface in the travel and tourism domain. The primary aim of this research work is to develop a human-computer interface for effective and efficient communication. NLP techniques ASR, POS, TTS has been combined with query processing techniques to retrieve the information. The user's perspective to the interface has been recorded, and response to the usability of the voice-based interface is observed. While attaining these goals, the focus was towards achieving high accuracy of speech processing engine along with efficient modeling of the natural language to generate SQL query. The end to end system performs with an accuracy of 94%. The purpose of this system development is to move a step ahead to empower the citizens of India digitally. The integration of the TTS engine extends the use of the system for visually impaired people.

The Impact of Virtual Human Voice on Learner Trust

The current study investigates if a virtual human’s voice can impact the user’s trust in interacting with the virtual human in a learning setting. It was hypothesized that trust is a malleable factor impacted by the quality of the virtual human’s voice. A randomized alternative treatments design with a pretest placed participants in either a low-quality Text-to-Speech (TTS) engine female voice (Microsoft speech engine), a high-quality TTS engine female voice (Neospeech voice engine), or a human voice (native female English speaker) condition. All three treatments were paired with the same female virtual human. Assessments for the study included a self-report pretest on knowledge of meteorology, which occurred before viewing the instructional video, and a measure of system trust. The current study found that voice type impacts a user’s trust ratings, with the human voice resulting in higher ratings compared to the two synthetic voices.

Implementation of Voice Controlled Virtual Assistant using STT, TTS and Logic Handling Engine

Implementation of Voice Controlled Virtual Assistant using STT, TTS and Logic Handling Engine

Development of TTS Engine for Indian Accent using Modified HMM Algorithm

A text-to-speech (TTS) system converts normal language text into speech. An intelligent text-to-speech program allows people with visual impairments or reading disabilities, to listen to written works on a home computer. Many computer operating systems and day to day software applications like Adobe Reader have included text-to-speech systems. This paper is presented to show that how HMM can be used as a tool to convert text to speech.

A Text-To-Speech Synthesis for Marathi Language using Festival and Festvox

This research paper describes the Impalement of the first, usable, Marathi Text to Speech system for Maharashtra Marathi using the open source Festival TTS engine. Besides that, this research paper also discusses a few practical applications that use this system. This system is developed using di-phone concatenation approach in its waveform generation phase. Construction of a di-phone database and implementation of the natural language processing modules are described. Natural language processing modules include text processing, tokenizing and grapheme to phoneme (G2P) conversion that were written in Festival's format. Finally, a test was conducted to evaluate the intelligibility of the synthesized speech.

The Prosodic Conditions in Robot's TTS for Children as Beginners in English Learning

A robot’s recognition and diagnosis of pronunciation and its speech are the most important interactions in RALL. This paper analyzes para verbal characteristics, F0 range and speaking rate, in English TTS engine speech which falls in the category of robot speech for Korean learners as beginners. In the actual result of this research, beginners did not change before and after repeating robot TTS, however it was confirmed that the speaking rate as the other prosodic variable increased close to the robot TTS’s speaking speed. Therefore robot TTS can be a suitable educational means to correct errors and train for speaking rate of English learners. However, it is still inadequate for the average F0 range of native speakers.

A Voice Calculator Based on Speech Error Correction

This paper presents our recent work towards the development of a voice calculator based on speech error correction and natural language processing. The calculator enhances the accuracy of speech recognition by classifying and summarizing recognition errors on numerical calculation speech recognition area, then constructing Pinyin-text-mapping library and replacement rules, and combing priority correction mechanism and memory correction mechanism of Pinyin-text-mapping. For the expression after correctly recognizing, the calculator uses recursive-descent parsing algorithm and synthesized attribute computing algorithm to calculate the final result and output the result using TTS engine. The implementation of this voice calculator makes a calculator more humane and intelligent.

안드로이드 OS 기반 한국어 TTS 서비스의 설계 및 구현

국내에서 출시된 안드로이드 기반의 스마트폰은 한국어 TTS 엔진이 내장되어 있지 않고, 구글에서도 공식적인 한국어 TTS 기술 개발을 발표하지 않고 있는 상황이다. 따라서 안드로이드 스마트폰을 사용하는 어플리케이션 개발자 및 사용자들의 불편이 갈수록 심해져 가고 있다. 본 논문은 안드로이드 기반의 스마트폰에서 서비스할 수 있는 TTS시스템의 설계 및 구현에 대해 기술하였다. 신속 명료한 TTS를 위해 안드로이드 NDK를 이용하여 텍스트 전처리와 합성음 생성 라이브러리를 구현하였다. 또한, 자바의 스레드 기법과 스트림을 적용한 AudioTrack 클래스 객체를 사용하여 TTS 응답시간을 최소화 하였다. 구현된 한국어 TTS 서비스를 테스트하기 위해 수신된 문자메시지를 읽어주는 어플리케이션을 설계 및 개발하였다. 평가 결과, 임의의 문장에 대해 자연스러운 합성음을 생성하였으며, 실시간 청취가 가능하였다. 또한, 어플리케이션 개발자들은 구현된 한국어 TTS 서비스를 이용하여 음성을 통한 정보 전달을 손쉽게 적용할 수 있다. 본 논문에서 구현한 한국어 TTS 서비스는 기존 제한적 음성합성 방식의 어플리케이션의 단점을 개선하였으며, 음성을 통한 정보전달 어플리케이션 개발자 및 사용자들에게 사용성과 편의성을 제공할 수 있다. Though Android-based smart phones are being released in Korea, Korean TTS engine is not built on them and Google has not announced service or software developer's kit related to Korean TTS officially. Thus, application developers who want to include Korean TTS capability in their application have difficulties. In this paper, we design and implement Android OS-based Korean TTS system and service. For speed, text preprocessing and synthesis libraries are implemented using Android NDK. By using Java's thread mechanism and the AudioTrack class, the response time of TTS is minimized. For the test of implemented service, an application that reads incoming SMS is developed. The test shows that synthesized speech are generated in real-time for random sentences. By using the implemented Korean TTS service, Android application developers can transmit information easily through voice. Korean TTS service proposed and implemented in this paper overcomes shortcomings of the existing restrictive synthesis methods and provides the benefit for application developers and users.

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.

The IBM expressive text-to-speech synthesis system for American English

Expressive text-to-speech (TTS) synthesis should contribute to the pleasantness, intelligibility, and speed of speech-based human-machine interactions which use TTS. We describe a TTS engine which can be directed, via text markup, to use a variety of expressive styles, here, questioning, contrastive emphasis, and conveying good and bad news. Differences in these styles lead us to investigate two approaches for expressive TTS, a "corpus-driven" and a "prosodic-phonology" approach. Each speaker records 11 h (excluding silences) of "neutral" sentences. In the corpus-driven approach, the speaker also records 1-h corpora in each expressive style; these segments are tagged by style for use during search, and decision trees for determining f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0</sub> contours and timing are trained separately for each of the neutral and expressive corpora. In the prosodic-phonology approach, rules translating certain expressive markup elements to tones and break indices (ToBI) are manually determined, and the ToBI elements are used in single f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0 </sub> and duration trees for all expressions. Tests show that listeners identify synthesis in particular styles ranging from 70% correctly for "conveying bad news" to 85% for "yes-no questions". Further improvements are demonstrated through the use of speaker-pooled f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0</sub> and duration models

Method and system for delivering text-to-speech in a real time telephony environment

A method and system for communicating information to a telephony user in response to a request for the information from the telephony user includes a text data source having text documents and a voice application operable with the text data source for retrieving a text document related to the requested information. A text-to-speech (TTS) manager divides the text document into text document segments. TTS engines convert the text document segments into speech segments. Each text document segment is converted into a speech segment by one TTS engine. A buffer reassembles the speech segments in the proper order and then delivers the speech segments to the user. The TTS manager divides the text document as a function of the rate at which speech segments are delivered to the user such that the speech segments are delivered from the TTS engines to the buffer and from the buffer to the user continuously.