Research & development

Allvoice Developments has extensive know-how within the speech and voice recognition marketplaces. The company prides itself on developing solutions that meet the real needs of customers and their customers. It also provides consultancy services for those organisations looking at adopting recognition technology.

SPEECH RECOGNITION . . . for Computer Systems

The founder and owner of Allvoice, John Mitchell, first looked at speech recognition coming out of the United States and thought that the dictation-to-screen products of that era were not very user friendly. Then, in the mid 1990's, the most likely users did not have a computer on their desk but dictated letters and reports blind onto a dictation machine. A skilled typist would then not only type up the document but would add functionality, such as, adding the full name and address of the recipient and generally creating the layout of the document - none of which the author would actually dictate, and may not comprehend how it was achieved using a computer's word processor. Those speech products expected the author to know how to do all this, and have the client's address readily available. In addition, if the author dictated non dictionary words, these would not be recognised by the speech recognition product. As a name and words in an address often fell into this gap, creating a letter with the early speech recognition products was a complex and awkward process, so the adoption of those products was limited, although many tried.

The speech recognition companies sought to fix their limited marketplace acceptance with incremental improvements in recognition accuracy. It was a matter of failing to see the forest through the trees. John Mitchell, and the team that he had assembled at that time, focused on the daily tasks that the end-users -- the author and typist -- needed to accomplish. Then, they developed features that made it easier for the author and typist to perform their individual tasks using speech recognition, and they did so in harmony with the traditional workflow between those users. Today, the latest products incorporate almost all of the features that were developed by Mr Mitchell and his team, some of which were patented and eventually licensed to many companies active in this field. The patents remain available for licensing to others, and new companies in the market as well.

SPEECH RECOGNITION . . . for Telephony Systems

Mr. Mitchell and his team built upon their success with computer systems by also creating innovative speech recognition products for telephonic systems. The early speech products in that field, as with speech products for computers, assumed that users would instantly accept the alternative of voice recognition automation to the mind numbing alternative of key pad data entry, especially when faced with a maze of pre-recorded menus that were often poorly designed or, in some instances, aimed at ensuring users could not get through to who they really needed to talk to, e.g., someone with knowledge and authority. Mr. Mitchell and his team worked on adding speech recognition features that would allow users to say what they wanted, as if an intelligent human had answered their call. In addition, they developed software that would recognise unknown callers using simple questions to save the laborious process of taking down name and address details prior to transferring the call. Again, much of the foresight that Mr Mitchell and his team put into practise has been adopted by many others.

For any enquiries regarding the above, please email in the first instance to