Website design
by
Accudata Systems

Continuous speech recognition software, to be useful and accurate, must build speech models specific to the language and subject matter of the speaker. Scanning documents, Dragon System's "Vocabulary Builder" augments its general vocabulary with the users specific terminology and the context in which it is used. This function, which takes place during the initial program installation and is repeatable at any time, increases accuracy up to 5% by customizing a vocabulary model specific to the user. This is extremely helpful in a general business or professional environment where specific terminology and context can only be derived from existing documents. However, this vocabulary (also called a topic or a context), does not satisfy the needs of legal and medical professionals using a large number of specialized words not appearing in the scanned documents during the vocabulary building process. Better and faster results can be achieved by integrating specialized vocabularies, skillfully crafted by professional teams of linguists, transcriptionists, and speech recognition experts.

To accomplish the task of constructing a specific vocabulary (i.e.medical or legal), there are 4 key elements that the vocabulary developers must deal with:

• The Acoustical Model, which incorporates the spoken sounds in the specific language.
• The Vocabulary Model, which is a list of words used for a particular subject matter with their spellings and pronunciations.
• The Language Model which consists of statistical information about the usage of each word alone and in combination with other words. For example, the language model may include the number of times and in what combinations "For," "4," "Four," and "Fore" would occur in a fairly large body of text.
• The Command Set, consisting of commonly used medical terms such as "blood pressure" which will correctly format the printed dictation. i.e. 120 over 80 prints 120/80.

A specialized vocabulary must meet its users full terminology needs. For example, a vocabulary for cardiologists must support dictating reports, including anatomical and procedural words. However, much of what a cardiologist dictates in discharge summaries and other reports is outside his/her specialty, so a cardiology vocabulary should also include a full internal medicine vocabulary as well. This same concept applies to all but the most general subject matters. Dragon's medical vocabularies feature 60+ medical specialities (see below for partial listing) including General Practice.

To assist its user in producing professional quality letters and reports, a specialized vocabulary must support the user style. When medical transcriptionist take dictation, they must ensure that capitalization, abbreviations, hyphenation, acronyms, and eponyms conform to their professional organizations style guidelines. A speech recognition vocabulary should follow the same guidelines.

Related vocabularies should consistently follow the same guidelines. For example, an internist, cardiologist, and general surgeon in the same practice will want to produce stylistically consistent reports and letters. Additionally, a cardiology vocabulary should be built on a strong internal medicine base and support the dictation of cardiac catheterization and EKG interpretation reports.

Continuous speech recognition programs such as the Dragon NaturallySpeaking Professional series is designed to achieve high recognition accuracy right out of the box. However, this accuracy will greatly suffer if the vocabulary is not robust and efficient enough for professional users such as physicians and attorneys. Accudata Systems supports the use of specialized vocabularies developed in collaboration with leading medical centers.

back home