IntelliSpeech pronunciation training

The voice recognition software IntelliSpeech is used during self instruction for intensive pronunciation training. Practicing pronunciation is the one thing that classroom tuition often neglects - partly due to social pressure but also to the fact that 45 minutes only leave so much time for every student to talk. Read here about dp's exclusively designed linguistic technology which coaches students in the pronunciation and intonation of words, phrases and entire sentences - and even allows students to control the learning program by speaking instructions into the microphone.

IntelliSpeech speech recognition is the dp trademarked feature that analyzes the exact phonemes used in speech, and provides the learner with an immediate evaluation of speech at several levels. Using a microphone connected to the pc students records directly into the CLT learning software. They can see immediately how well the spoken text corresponds with the pre-recorded speakers and phoneme database by seeing blue bars which appear over the spoken sentence. On another level, the student can see a graphic depiction of how his spoken text best represents the intonation of the spoken sentence. The student can replay the original sentence or replay his own voice for comparison purposes.



A scale is provided showing how closely the student’s pronunciation has come to corresponding to the pre-recorded samples. And if the student has problems with a specific word, he can practice that individual word by simply clicking on the word to switch into single word mode. The results of the student 's work on pronunciation are then stored in the CLT LMS for further analysis by the personal online tutor.

How IntelliSpeech works
IntelliSpeech was designed to help the student's pronunciation refinement and to approach that of a native speaker. IntelliSpeech shifts the focus away from trying to determine whether a word was spoken, to evaluating how a word was spoken. IntelliSpeech not only uses a phonemic database but also a character feature in order to determine how well something was said in comparison to real speech. With IntelliSpeech, the learner can focus on perfecting speech through multiple comparisons with numerours male and female speakers. Every word which is used in the program has been recorded using a multitude of different speakers; once an utterance has been made the program compares it to the phoneme databank. In this way, from the first attempt, the learner receives professional correction, even if the word or sentence is spoken slowly, with pauses or hesitation. The result: The closer the learner's pronunciation is to that of the native speakers', the higher the score. Unique is how, like a teacher, the program evaluates every word in the sentence in isolation

Why did dp develop its own speech recognition software?
Although a natural speech interface is more commonly used in contemporary software applications, it is usually not optimized for language learning. Speech recognition has been mainly designed to facilitate the use of software, not to improve pronunciation. Common speech recognition software simply breaks down utterances into phonemes. These units are then compared with phonemes found in the program's database and the software tries to determine a probable match. The pitfall of this approach in language learning software is obvious: it does not evaluate the quality of speech input. Quite to the contrary, the more unnatural the pronunciation, the easier the recognition. IntelliSpeech, designed on site by dp, closes this gap - the user can simultaneously control the program and improve pronunciation.

Where IntelliSpeech is used in CLT
The Voice Pilot 'understands' the students and almost like a real teacher. Plus: You can actually control the application by spoken commands like "Tim, correct this exercise please"!

Upon completing an exercise in which pronunciation is practised, the IntelliSpeech panel appears in the upper part of the screen under the video tutor. The learner can click on any sentence, listen to it being spoken and repeat it. The program corrects every single word within the sentence and graphically displays how well the sentenced was pronunced - an absolute language learning novelty.

In the interactive dialogs the learner can take on a participative speaking role, which, with correct pronunciation, is almost like a real conversation. The program automatically jumps to the next scene as soon as the IntelliSpeech accepts the learners pronunciation as being good enough. In the interactive dialogs the user has the possibility not only to practise complete sentences, but also a natural conversation.

Many exercises in the interactive language tour use the Speak & Drop technique or voice input to solve them. Whenever a small microphone symbol appears in an answer gap, the student simply has to speak the answer into the microphone without touching the keyboard or mouse.

return