Speech Recognition and Text to Speech (TTS)

Automatic Speech Recognition

SmartSpeech is an integrated platform for ASR (Automatic Speech Recognition ), TTS (Text To Speech) and supports Microsoft and ScanSoft engines. ASR makes IVR speech enabled and helps the caller to interact with the system in a convenient and natural way. TTS is used to convert the Text data to Speech on the fly.

SmartSpeech has a two tier architectural deployment with the Resource Manager and the Speech Engine can be either SAPI, ScanSoft or any other standard Speech Engine. The Resource Manager handles queuing of audio stream inputs to the Speech Engine and the same performs the recognition through the grammar and the audio preprocessing layers.

SmartSpeech offers a robust platform with which voice barge-in is possible. The caller can cut through the Voice menu for a specific Voice prompt. With continuous speech processing, the incoming voice data from the caller is processed using the DSP-based Echo Cancellation and Voice Activity Detection integrated on the voice card. For example, in an area where the user is familiar with the speech flow of the system, he might not want to go through all its voice prompts. He would instead, directly want to reach a specific prompt to seek the specific information.

SmartSpeech can be deployed with the voice enabled application to function in any environment like workstation or client/server network. Incase of a single workstation setup, SmartSpeech and the voice enabled application will require windows messaging system. Incase of the client/servernetwork setup, there would be multiple clients in which SmartSpeech application will be loaded. The voice enabled application/IVR will interact with these clients to get the recognition output.


  • Offers multiple grammar support to handle large vocabulary.
  • Supports Real Time Learning (RTL) to tune the User Profile for recognition precision.
  • Speaker Independent.
  • Recognition accuracy above 90%.
  • Allows communication using the Windows Messaging System.
  • Offers easy integration with any other voice enabled applications.
  • Natural sounding speech in Indian, American and British accent output.

Text to Speech

Bay TalkiTec's SmartSpeech has built-in Text-To-Speech engine which is a state-of-the-art solution designed for integration into voice processing platforms and speech-enabled systems. SmartCall supports Microsoft SAPI and ScanSoft Text-To-Speech engines.



Intelligent pre-processor correctly interprets e-mails, URLs, excludes footers, filters, headers, expands abbreviations and acronyms, and translates emotions.


Unlimited vocabulary results from pronunciations generated by phonetic rule, rather than looked up in a dictionary or table.

Ability To Speak Proper Names

Advanced Transliteration enables pronunciation of names in the right manner they have to be spoken. Understands, renders and speaks names that are difficult to be pronounced.

Spell Check

Checks for errors in spellings of text input and alters if any. Has a large dictionary of words from which it picks the nearest match to the erroneous word and replaces the wrong one with the nearest match found.

Speech Variations

Various accents and rules aside, the engine also supports variable speed at which speech is delivered. The pitch and the volume of speech as well as introducing pauses while speaking the given text can be fine tuned with the aid of TTS tool.


  • News on Demand
  • E-Mail Reader
  • Directory Enquiry
  • Talking Yellow Pages
  • Voice Portal

Latest News