Hindi Text-to-Speech (TTS) System

The Text-to-Speech System developed by the GIST group caters at present to Hindi and offers three different Text to Speech synthesis systems, classified as per the needs of the Industry.

  • On the one hand a LIMITED DOMAIN TTS SYSTEM geared to the requirements of systems where the number of sentences to be spoken is limited.

  • A GENERIC TTS SYSTEM which is a hybrid model (diphones + recorded words) and which can be used as a web-service to be used for a large number of web-applications where there is a need for voice rendering.

  • Finally a TINY TTS SYSTEM based on the phoneme level provides a library of a tiny phoneme set which can be deployed on hand-held devices like mobiles, PDAs etc.

With few modified rules system can be adapted for other Indian Languages also.

TTS Architecture :

The TTS system has a sophisticated architecture with a robust NLP (Natural Language Processing) back-end and a high quality Natural Voice system at the front end.

 

Main Text-to-Speech System Features :
The main features of GIST TTS which distinguishes it from all other existing systems are as under :

1. Strong heuristic

  • Naturalness of speech quality in all three systems.
  • A strong linguistic back-end driven by statistical analysis which can handle all inherent problems of TTS in Hindi such as Schwa deletion and Syllable identification.
  • Robustness of the Engine
  • Language independent routines both for the linguistic analysis as well as for the synthesis front-end enabling thereby the creation of TTS in other Indian languages
  • Formatted Numbers Conversion like Date, Time, Float, Currency, Ordinal, Expression etc. from the running text.

2. Economical in memory

  • Low system requirements (P-III, 64mb RAM)
  • Low size Database (3mb, 6k-mono wav)

3. Input and Output

  • Supports Unicode and ISCII
  • Phonetic keyboard eg. typing kar gives कर
  • Standard Inscript Keyboard
  • Conversion of input Text to WAV, AMR or MP3 file format.

4. Robust Acronym and Number to Word routines : The best in the industry

  • Acronym expansion, Look-up Dictionary to handle commonly used words.
  • Identifies numbers in Date, Float, Time, Currency, Ordinals forms and expands those in proper manner.
  • A robust Number to Word routine which deploys heuristic rules for distinguishing the different possibilities of Num toWord
  • Converts Number to word and then into the speech format.
  • On the fly translation of number in 19 different Indian languages.
  • Supports maximum possible range of numbers.
  • Reads Number in two forms : digit as well as in number format.
  • Speaks out 42 digit number in unit form as well as of any length in digit form.
  • User can select unit range for numbers.

5. Supported platforms

  • Windows
  • Linux
  • ARM Platform

6. Prosody and Intonation : Inclusion of Prosody and Intonation are in process to produce more natural speech and this feature will be implemented at the earliest.

Text-to-Speech Applications :

Hindi TTS System on windows Mobile 5.0 to speak messages.

Voice Reminder
You can set reminder for meetings, anniversary or To Do calls using Voice Reminder application. On event time, Voice Reminder will speak your note.

Calendar Set Reminder
   
Voice Reminder reading out message for you.
   

 

Click below for Audio . . .

For more details, please contact:

More information on GIST products
E-Mail:
info.gist@cdac.in

Sales related information
E-Mail:
sales.gist@cdac.in

Support related information
E-Mail:
support.gist@cdac.in