Sign in Subscribe
Pharmaceutical industry

Revolutionizing Computer Interaction: IBM's VoiceType Dictation Speech Recognition System

Strategic Insights

— 2 min read

IBM's latest innovation in speech recognition, VoiceType Dictation, has the potential to transform the way people interact with their computers. This cutting-edge technology allows users to bypass the keyboard and mouse by talking directly to the PC, providing a game-changing solution for individuals with disabilities, those suffering from repetitive strain injury, and those who require an additional hand while working on their computers. With a vocabulary of up to 32,000 words and available in six languages, VoiceType Dictation is an advanced tool that has the capability to accurately transcribe user speech into text.

Key Takeaways:

  • VoiceType Dictation has a vocabulary of up to 32,000 words and is available in six languages, including English.
  • The system uses a plug-in card and a microphone headset to capture user speech, which is then broken down into phonemes and compared against the software's vocabulary to find the closest matches.
  • The system can accurately transcribe up to 70-100 words per minute, with an 86% accuracy rate for a particular user, which improves rapidly with use.
  • Users can define up to 2,000 terms specific to their type of work, or purchase additional vocabularies like medical and legal.
  • The system uses statistics to determine the likelihood of each word making contextual sense and can substitute words on the reserve list if necessary.
  • A grammar checker is also included to differentiate between words that sound the same, such as 'too', 'two', and 'to'.
  • Users can issue commands and controls to the PC using VoiceType, such as cutting and pasting a document from VoiceType to WordPerfect.
  • Controller cards for VoiceType Dictation can be supplied to fit standard ISA and PCMCIA slots in the computer, as well as IBM's own MCA.
  • Versions are available now for the OS/2 operating system, with a US English version for Windows expected to be available soon and European language versions for Windows expected in February.

Statistics:

  • The system can accurately transcribe up to 70-100 words per minute.
  • The system has an 86% accuracy rate for a particular user, which improves rapidly with use.
  • Users can define up to 2,000 terms specific to their type of work.
  • The system requires at least 12MB main memory and 60MB hard disk during installation.
  • The minimum processor required for the OS/2 version is an 80486SX, while for Windows it will require the more powerful 80486Dx processor.

Sources:

  • "FOR those busy types who sometimes feel the need for a third, or even a fourth, hand while using a personal computer, the development of speech recognition systems could prove a saviour." - Original text, no date provided.
  • IBM's VoiceType Dictation software, available in six languages.
  • IBM's VoiceType Dictation system, available for OS/2 and Windows operating systems.

Read more