Advanced Hands Free Computing

Discourse acknowledgment innovation is now accessible to Advanced education and Further Training as are a considerable lot of the contrasting options to a mouse. In this task, we have proposed another application for hands-free registering which utilizes voice as a noteworthy correspondence intend to help the client in checking and processing reason on his machine. In our undertaking as we have for the most part utilized voice as correspondence mean. Discourse innovation envelops two advancements: Discourse Acknowledgment and Discourse Blend. In this venture, we have specifically utilized discourse motor which utilizes Shrouded Markov Model and Highlights extraction strategy as Mel scaled recurrence cepstral.

The meal scaled recurrence cepstral coefficients (MFCCs) got from Fourier change and channel bank examination are maybe the most broadly utilized front closures in best in class discourse acknowledgment frameworks. Our point is to make an ever-increasing number of functionalities which can enable human to aid their day by day life and furthermore to lessen their endeavors. The Well (Concealed Marcov Display) is utilized inside in which the state isn't specifically unmistakable, yet yield, reliant on the state, is obvious. Each state has a likelihood of dissemination over the conceivable yield tokens. In this manner, the arrangement of tokens produced by a Well gives some data about the grouping of states.

Research in discourse preparing and correspondence generally, was propelled by peoples want to fabricate mechanical models to imitate human verbal correspondence abilities. Discourse is the most normal type of human correspondence and discourse preparing has been a standout amongst the most energizing zones of the flag handling. Discourse acknowledgment innovation has made it workable for PC to take after human voice summons and comprehend human dialects. The primary objective of discourse acknowledgment territory is to create procedures and frameworks for discourse contribution to the machine.

There are various handicaps and therapeutic conditions that can bring about obstructions for those endeavoring to utilize a standard PC console or mouse. This does not simply incorporate physical handicaps. Numerous understudies with perusing/composing troubles, for example, dyslexia can discover utilizing the console to enter content into the PC a relentless exercise that can confine their inventiveness. Sans hands registering is a term used to portray an arrangement of PCs so they can be utilized by people without the utilization of the hands interfacing with ordinarily utilized human interface gadgets, for example, the mouse and console.

This application fundamentally joins two advances: Discourse union and Discourse acknowledgment. Through Voice Control, the PC utilizes voice prompts to ask for a contribution from the administrator. The administrator is permitted to enter information and to control the product stream by voice summon or from the console or mouse. The Voice Control framework takes into consideration dynamic determination of a language structure set or legitimate arrangement of summons. The utilization of a diminished sentence structure set extraordinarily builds acknowledgment precision. Discourse Acknowledgment (otherwise called programmed discourse acknowledgment or PC discourse acknowledgment) changes over talked words to content. Discourse Acknowledgment takes a sound stream as information and transforms it into a Charge which is later mapped with an occasion. Discourse blend content is changed over to discourse flag. Discourse blend is otherwise called content to discourse transformation.

In this application, discourse blend is utilized to peruse mail and for changing over content into the discourse. In our task, we have utilized The Discourse Application Programming Interface or SAPI. It is a Programming interface created by Microsoft to permit the utilization of discourse acknowledgment and discourse blend inside Windows applications. As a rule the sum total of what Programming interface have been composed with the end goal that a product engineer can compose an application to perform discourse acknowledgment and amalgamation by utilizing a standard arrangement of interfaces, open from an assortment of programming dialects.

Moreover, it is feasible for an outsider organization to deliver their own Discourse Acknowledgment and Content To-Discourse motors or adjust existing motors to work with SAPI. Fundamentally Discourse stage comprised of an application runtime that gives discourse usefulness, an Application Program Interface (Programming interface) for dealing with the runtime and Runtime Dialects that empower discourse acknowledgment and discourse blend (content to speech or TTS) in particular dialects.

Benefits of using system speech:

1. Microsoft .NET Framework Managed-Code APIs

2. Speech Recognition

3. Speech Synthesis (text-to-speech or TTS)

4. Standards Compatible

5. Cost Efficient.

Limitations in the existing system

Noises, distortions, and unforeseen speakers seldom cause difficulty for a human to understand speech signals whereas they seriously degrade performances of automatic speech recognition (ASR) systems. While extracting features from speech, it becomes difficult to recognize correct word due to noise and other environmental conditions. Windows speech recognition is efficient but it is like one-way communication. When words are spoken, processing is done and the reply is given by performing a task or opening application. It is hardware or software response instead of voice. It is necessary to get voice feedback for the command given by the user for any user-friendly application.

In Window Speech API only OS related commands are executed. These commands are helpful, but they are not commanded to assist in user life to make their life easier. This project adds commands for making device handier. All commands which can be executed by command prompt are included. Windows speech API does not contain hardware commands. We can open Google by voice command but we can?t type our query by voice.

Also there are number of limitations like environment issues due to type of noise, signal/noise ratio, working conditions, transducer issues, channel issues due to Band amplitude, distortion, echo etc., speakers issues due to Speaker dependence/independence, Sex, Age, physical and psychological state, speech style issues due to voice tone(quiet, normal, shouted) etc., production issues due to isolated words or continuous, speech read or spontaneous speech speed(slow, normal, fast), vocabulary issues due to Characteristics of available training data, specific or generic vocabulary and many more which limit the efficiency application.

Proposed system

Speech recognition process can be completed in two parts - front end and a back end. The front end processes the audio stream, isolating segments of sound that are probably speech and converting them into a series of numeric values that characterize the vocal sounds in the signal. The back end is a specialized search engine that takes the output produced by the front end and searches across three databases.

As user gives the speech signal (simply an audio stream) with the help of a microphone. Microphone processes the audio stream to the Speech Recognition system which will convert a speech signal to a sequence of words in form of digital data i.e. a command with the help of SAPI. This command is then searched in context database according to context search. If it matches then further action mapping is done in which actions or response to the specific command is specified. Using application interface APIs like keyboard events, mouse events, and OS interface, appropriate action is performed according to given command To perform this whole operation Speech recognition and synthesis is used which we are going to see in detail.

How speech recognition works

Speech recognition fundamentally functions as a pipeline that converts PCM (Pulse Code Modulation) digital audio from a sound card into recognized speech. The elements of the pipeline are as follows.

1) Transform the PCM Digital Audio

The digital audio is a stream of amplitudes, sampled at about 16,000 times per second. To make pattern recognition easier, the PCM digital audio is transformed into the "frequency domain." Transformations are done using a windowed fast-Fourier transform. The fast Fourier transform analyzes every 1/100th of a second and converts the audio data into the frequency domain. Each 1/100th of a second result is a graph of the amplitudes of frequency components, describing the sound heard for that 1/100th of a second. The speech recognizer has a database of several thousand such graphs (called a codebook) that identify different types of sounds the human voice can make. The sound is "identified" by matching it to its closest entry in the codebook, producing a number that describes the sound. This number is called the "feature number."

2) Figure Out Which Phonemes Are Spoken

In an ideal world, you could match each feature number to a phoneme. If a segment of audio resulted in feature #52, it could always mean that the user made an "h" sound. Feature #53 might be an "f" sound, etc. If this were true, it would be easy to figure out what phonemes the user spoke. Unfortunately, this does?t work because of a number of reasons. Every time a user speaks a word it sounds different. The background noise from the microphone and user?s office sometimes causes to recognize different feature number. The sound of a phoneme changes depending on what phonemes surround it. The "t" in "talk" sounds different than the "t" in "attack" and "mist". The background noise and variability problems are solved by allowing a feature number to be used by more than just one phoneme and using statistical models to figure out which phoneme is spoken.

3) Convert the Phonemes into Words

4) Reducing Computation and Increasing Accuracy

The speech recognizer can now identify what phonemes were spoken. Figuring out what words were spoken should be an easy task. If the user spoke the phonemes, "heh lone", then you know they spoke "hello". The recognizer should only have to do a comparison of all the phonemes against a lexicon of pronunciations.

5) Context Free Grammar

One of the techniques to reduce the computation and increase accuracy is called a "Context Free Grammar" (CFG). CFG?s work by limiting the vocabulary and syntax structure of speech recognition to only those words and sentences those are applicable to the application?s current state. The application specifies the vocabulary and syntax structure in a text file. The speech recognition gets the phonemes for each word by looking the word up in a lexicon. If the word is?t in the lexicon then it predicts the pronunciation.

6) Adaptation

Speech recognition system "adapt" to the user?s voice, vocabulary, and speaking style to improve accuracy. A system that has had time enough to adapt to an individual can have one fourth the error rate of a speaker independent system. The recognizer can adapt to the speaker?s voice and variations of phoneme pronunciations in a number of ways which are done by weighted averaging.

The resultant project operates for all control panel commands as well as all function keys, accelerator keys (combination of 2 or more shortcut keys). Gmail shortcuts can be executed. Many of the hardware commands can be operated. The user can see the recognized command in a textbox. Also, the user can dictate words in a text file using notepad. A feature named „Speech pad? is provided. Using speech pad user can read any text files wherever it is stored. This feature is completely voice operated. Voice chat can be done with PCs connected to each other.

This paper gives a brief idea about Hand Free Computing Application which will help disabled users by eliminating the use of keyboard and mouse in most of the applications. Likewise, disabled persons may find hands-free computing important in their everyday lives.