 |
Teams and Tasks
The technical resources are divided into four major teams:
1. Data Collection and Tagging Team
This team consists of linguists who are working on the tasks of data collection and tagging. Each linguist works on different phases of speech corpus design such as the creation of word lists for recordings, cleaning of recorded data and data tagging for TTS and ASR technologies. The main challenges include the speech corpus design, recordings and management of data. Another challenge is the manual annotation of data which is a time consuming task. The high accuracy of annotation will ensure that the TTS and the ASR engines will generate high quality speech and low word error rate respectively.
2. Telephony and Dialog Framework Team
This team works on design and development of telephony and dialog frameworks. The resources have setup the telephony framework which is necessary for creating a channel of communication between the users and the dialog system. Moreover, a dialog framework is developed that generates seamless communication among the different modules of the dialog system and to provide assisted services to the users. The main challenges include the selection and setting up of telephony framework, development and optimization of the dialog system architecture.
3. Text-to-Speech Synthesis Team
This team consists of computational linguists and developers. They are working on the development of Urdu TTS system. The TTS system will be integrated within the dialog system and the screen reader. The main challenges include the development of different utilities for data preparation, building TTS system on state-of-the-art technologies such as unit- selection algorithm and Hidden Markov Models (HMM), testing and improving the voice quality of the TTS system. Another challenge includes the integration of TTS system within the dialog system and the screen reader.
4. Automatic Speech Recognition Team
This team consists of computational linguists and developers. They are working on development of Urdu ASR system for weather and location domains. The ASR system will be integrated within the dialog system which will provide weather information and location-based services. The main challenges include recording of speech data on mobile phones covering different accents in Pakistan, building ASR system for weather and location domains and integration of ASR system within the dialog system.
1. Data Collection and Tagging Team
This team consists of linguists who are working on the tasks of data collection and tagging. Each linguist works on different phases of speech corpus design such as the creation of word lists for recordings, cleaning of recorded data and data tagging for TTS and ASR technologies. The main challenges include the speech corpus design, recordings and management of data. Another challenge is the manual annotation of data which is a time consuming task. The high accuracy of annotation will ensure that the TTS and the ASR engines will generate high quality speech and low word error rate respectively.
2. Telephony and Dialog Framework Team
This team works on design and development of telephony and dialog frameworks. The resources have setup the telephony framework which is necessary for creating a channel of communication between the users and the dialog system. Moreover, a dialog framework is developed that generates seamless communication among the different modules of the dialog system and to provide assisted services to the users. The main challenges include the selection and setting up of telephony framework, development and optimization of the dialog system architecture.
3. Text-to-Speech Synthesis Team
This team consists of computational linguists and developers. They are working on the development of Urdu TTS system. The TTS system will be integrated within the dialog system and the screen reader. The main challenges include the development of different utilities for data preparation, building TTS system on state-of-the-art technologies such as unit- selection algorithm and Hidden Markov Models (HMM), testing and improving the voice quality of the TTS system. Another challenge includes the integration of TTS system within the dialog system and the screen reader.
4. Automatic Speech Recognition Team
This team consists of computational linguists and developers. They are working on development of Urdu ASR system for weather and location domains. The ASR system will be integrated within the dialog system which will provide weather information and location-based services. The main challenges include recording of speech data on mobile phones covering different accents in Pakistan, building ASR system for weather and location domains and integration of ASR system within the dialog system.
 |
042-36811113 |
 |