AN INSIGHT ON THE AUTOMATIC SPEECH RECOGNITION SOFTWARE

The art of speech and the art of transcription are blended to bring about a new state-of-the-art technology called as the automatic speech recognition software. The ASR or the automatic speech recognition software is found to be the talk of the town. Speech recognition has been a dream for us from the good old days of star wars and other science fiction movies and stories. Have our dreams come true?? Today, it has been partially fulfilled with the new arrivals in the markets. Each company has been into this competition of giving the best speech recognition software to the world market. What hasrnhappened to the race among themselves? It reminds me of the hare and the tortoise story. The slow and steady looks like it has won the race, but yet has miles to touch the finish line. Discussing about what exactly is the goal of the race?? Is it either getting to the top or getting to the people, is again a million dollar question. With all the revenues pooled in for speech recognition have started to drain, there is a need to analyze the growth with time factor, which will clearly show a flattened graph showing the stagnant nature of the software research and development.

Imagine a situation, where you have invested on speech recognition software for some thousand dollars per month and find it to be unworthy since they type in your dictations wrongly,words are replaced and jumbled, and the context becomes different, what a chaos that would create.The frustration that is exhibited at those times is really unbearable. Flawless products or services are nowhere to be found since everything on earth comes with unique pros and cons. This applies to the speech-to-text software as well. It has its own flaws and demerits, which limits the usage of it within the small community. The concept needs more attention and research to reach or to compete with the languages that have been developed over millions of years.

The ethnologue of the world seems to be far too long and unending. The languages that we speak today are the development of it over millions of years together with all the efforts of millions of generations. All animals communicate with each other, but it is only the humans who have formulated the communication in predefined set of signals known as the language. The Cortical Speech Center is again an evolutionary feature that only the humans posses, which differentiates the human brain from the other animals in the animal kingdom. Hence, the speech recognition softwares that has a very recent history compared to the languages has to travel not millions but at least few decades to understand the least about the speech and languages spoken by different groups of people.

The drawbacks of the voice recognition or audio-to-text software are:

• It cannot understand all the words after spending hours together training the software. Time is precious after all we have only 24 hours a day!!!

• All the punctuations such as coma, full stop, semicolon, hyphenation requires the speaker to dictate wherever he/she wants one.

• Understanding the context is another major drawback or demerit: Some words especially in English have many meanings and needs to be used in the correct context to obtain good results in the records. The software does not seem to understand the context in most of the places.

• Homophones are again a difficult task to handle for the audio to text software: Different words with the same pronunciation but different meanings: For example elicit-illicit; desert-dessert; there-their; flour-flower; bowel-bowl; words with same pronunciation but different spelling and meaning, which are used in different context, confuse the software resulting in bloopers and hilarious phrases and sentences.

• The other major black mark about the speech recognition is that it cannot understand the varied types of accent that is present in one single language. Understanding the words in a neutral slang itself is difficult for the software then how can it ever understand the different slangs or accents used by different people around the world!!

In 1997, Bill Gates gave a open statement that "In this 10-year time frame, I believe that we’ll not only be using the keyboard and the mouse to interact, but during that time we will have perfected speech recognition and speech output well enough that those will become a standard part of the interface." Now, it is 3 years past a decade and yet speech recognition is only at the primitive stage of usage and development.

Hence, to conclude transcription industry has a bigger hand over the audio-to-text software. Transcriptionists are not obsolete. They have their own space and need in the field for their integrity, caliber, and experience in the industry.