N/APosted on - 04/12/2012
How good is the current generation of voice recognition software compared to the "Dragon Naturally Speaking" program that I used to work on before? And has it reached to a point where it really can be used for dictation without spending a lot of time correcting the mistakes made by such obsolete program?
Current voice recognition vs. Dragon Naturally Speaking
The voice recognition software has changed considerably. At present, the reduced processing time and the high speed of implementation and high recognition value make these products the first choice of assistants.
You will be surprised by the fact that there is only two major unique applications in this field. This is explained by the fact that at present, the particular user will be limited by a particularly restricted. Indeed, if the professional field remains competitive, it is clear that he does not hold true for consumer applications. The abandonment of IBM ViaVoice on French soil leaves the field open to Dragon Naturally Speaking. Windows Vista saves us from this monopoly by incorporating a native voice recognition module.
However, to make the most of speech recognition software, it is essential to initiate and train thoroughly the user.
Moreover, it is possible to further increase the accuracy of speech recognition, and thus its effectiveness, by providing a basis for specialized vocabulary.
Also we need to consider that software is not able to distinguish speech sounds of ambient noise. For the system to work properly, it will be imperative to use a quality microphone (a headset comes with Dragon Naturally Speaking).
Another difficulty, not least, the French language (for example) has many homophones. This term refers to words or groups of words that sound pronunciation is identical. To avoid this pitfall, an elementary application of speech recognition uses the context of the sentence. Recognition will be more relevant with long sentences with small groups of words. The intonation will also play an important role. In conversation, the accent or tone of voice does not hinder understanding. Unlike humans, the programs are not able to overcome this pitfall. In most cases, we will have to go through a learning phase required.