top of page
  • Dragon Dictate

How does speech to text software work?

Speech to text software converts spoken words into written text. This process is also often called speech recognition or voice recognition. Although these terms are almost synonymous, speech recognition is sometimes used to describe the wider process of extracting meaning from speech, i.e. speech meaning. The term voice recognition is often associated to the process of identifying a person from their voice, i.e. speaker recognition. However, for all intents and purposes voice recognition, Speech Recognition and speech to text are the same thing. You speak and words appear in front of you on your computer. You will need to add punctuations such as “full stop” and “comma” as well as “new paragraph.” There are thousands of other voice commands which can be used with more sophisticated speech to text solutions such as Dragon Medical practice edition which is designed for medical doctors.

speech to text

How does speech to text work?

All speech to text systems rely on an acoustic model and a language model. Large vocabulary systems (such as Dragon Medical) use a pronunciation model. It is important to understand that there is no such thing as a universal speech recogniser. To get the highest speech to text quality, all of these models can be specialised for a given language, accent, application domain, type of speech, and intended specialty use such as medical, legal, business and other important fields.

Like any other digital recognition technology, speech to text systems will make mistakes. The speech recognition transcript accuracy is highly dependent on the author, the style of speech and the quality of the audio being recorded. Speech to text systems are far more complex than what most people think. Humans are used to understanding speech, not transcribing it, and only speech that is well dictated can be transcribed with 100% accuracy. For example you may be “uming”, “ahhing” , pausing and speaking rapidly within one paragraph without even realising this. Another human being may be able to fill in the gaps through their experience in listening to millions of minutes of audio over their life. But a computer speech-to-text system have to make sense of all these extra words and strange gaps in your dictation. It uses special algorithms and artificial intelligence to try and interpret your voice. However, this will never be 100% accurate because you will not dictate 100% accurately.

voice to text

User expectations are a big factor in speech to text systems. Some authors do not realise that their dictation is very difficult for a human to understand therefore almost impossible for a computer-based voice recognition system to interpret. Noting this, most speech to text users adapt their dictation style quickly after using speech recognition as they see the results of their poor dictation. They immediately start improving their dictation and this has a dramatic improvement in accuracy and responsiveness.

In other words speech to text system trains the speaker eventually to dictate better!

The net result is that within about a week or so most authors will be getting 98% accuracy plus out of their speech to text software.

There really is no point paying expensive transcription services to manually type your dictation. You are paying far too much money annual waiting too long to receive your dictation back.

Start dictating directly into your email, word documents, electronic health records, legal documentation, Internet web forms and just about any other environment where you need to type.

Stop waiting on someone else to do it and stop paying them to do it. Start using speech to text software today. Contact us to find out how or Call 1300 255 900 right now.


bottom of page