Communication plays a vital role in our lives. Humans started with signs, symbols and evolved to a stage, where they began communicating with each other using various languages. Then a paradigm shift happened and with the advent of computing and communication technologies, machines started communicating with humans and in some cases with themselves. This shift created the world of the internet, or as we technically know the Internet of Things(IoT) and gave rise to new ways of using data, where humans are able to communicate directly or indirectly with machines by training them which is known as Machine Learning. Previously, a person has to access a computational device in order to communicate with machines. But the extensive research and development in this area have eliminated the use of a computational device to a great extent as a medium of communication between humans and machines. This giant leap in communication is known as Automatic Speech Recognition and is based on natural language processing, that allows humans to interact with machines using their natural language in which they speak.
The preliminary research and development in the field of Speech Recognition have been successful, and now the speech scientists and technologists aim to correctly optimize the speech recognition engines according to the situations in which the machines communicate so as to reduce error rates and for improved efficiency. Some companies in the IT industry started spreading their roots in the development of speech recognition technologies. From more than a decade McLean, Virginia based organization GoVivace Inc. is continually specializing in design and development of speech recognition technologies and solutions. The company provides a wide range of products and solutions based on Speech Technology like voice biometrics, speech to text software (audio transcription), call analytics solutions(CALLai) and real-time captioning solutions(captionAI).
The Automatic Speech Recognition technology is a combination of two different branches – Computer Science and Linguistics. Computer Science to design algorithms and to program and Linguistics to create a dictionary of words, sentences, and phrases.
The first stage of development happens with speech transcriptions, where the audio is manually converted into text i.e speech to text conversion. After conversion, the software tries to remove unnecessary signals or noise by filtering the signals. Since humans talk at different speeds while uttering a word or sentences the generative model of speech recognition is designed to account for those rate changes. Later the signals are further divided to identify phonemes, the letters that have the same level of airflow, like ‘b’ and ‘p’. After identifying the phonemes, the program tries to match the exact word by making a comparison with the words and sentences that are stored in the linguistics dictionary. The speech recognition algorithm uses statistical and mathematical modeling to determine the exact word. Speech recognition software is of 2 types, one with learning mode and other as a human dependent system.
With the developments in Artificial Intelligence and the Big Data, speech recognition technology achieved the next level. A specific neural architecture called long short – term memory, bought a significant improvement in this field. Globally various organizations are leveraging the power of speech in their premises at different levels for a wide variety of tasks. For instance, the speech to text software can be used for converting audio files to text files with timestamps and confidence score for each word. Many countries do not have their language embedded keyboards, and a majority of people do not have an idea of using specific language keyboard, though they are verbally good at it. In these cases, speech transcriptions help them to convert speech into text in any language by hearing the speaker’s voice.
The other use of this technology is in real-time. This is also called as Computer Assisted Real-Time translation. This is basically a speech to text system which operates on a real-time basis. Organizations all over the world perform meetings and conferences, so for maximum participation by the global audiences, they leverage the power of live captioning systems i.e. captionAI. The real-time captioning system converts the speech to text and displays it on the output screen, translate the speech in one language to the text of other languages and also help in making notes of a presentation or a speech. The system can convert speech to text that is also understood by hearing-impaired people.
Apart from speech to text, speech recognition technology spreads its branch into the biometric system, which created voice biometrics for authentication of users. Voice biometric systems analyze the voice of the speaker which depends on factors like modulation, pronunciations, and other elements. In these systems, the sample voice of the speaker is analyzed and stored as a template. Whenever the user utters the phrase or sentence, the voice biometrics system compares them with the stored template and provides authentication. However, these systems faced a lot of challenges. The human voice is always affected by physical ailment or emotional state of a person. The recent advancements in voice biometric systems operate by matching the phrase with the sample and started analyzing the voice patterns by taking psychological and behavioral voice signal into consideration. Also, the advance developments in voice biometrics technology are going to benefit financial institutions, banks or enterprises where data security is a major concern.
Analytics also played a significant role in the development of speech recognition technology. Big data analysis created a need for storing voice data. Call centers started using the recorded calls for training their employees. Since customer satisfaction is now the major focus of organizations around the globe. Now, organizations want to track and analyze the conversation between executives and customers. And to ease off this vigorous task GoVivace Inc. has developed a call analytics solutions i.e. CALLai, which monitors and measures the performance and analytics of call. This call analytical solution enhances the performance of services provided by call centers. Through this one can classify their customers and can serve them better by giving faster and favorable responses.
Way Ahead For Speech Recognition Technology
Research in speech recognition technology has a long way to go. Until now, the software can act on instructions only. The human communication feel does not exist with those machines. Researchers are trying to inculcate the human responsiveness into machines. They have a long way to go in the innovation of speech recognition technology, and human language technology. The primary aspect of research concentrates on, how to make speech recognition technology more accurate. More Accuracy is required in understanding the human language. For example, a person raised a question “how do I change camera light settings?” which technically means that he wants to adjust the camera flash. So major concentration is on understanding the free form language of humans before answering them.
Speech recognition technology is already made its way into the organizations and started providing effective and efficient results. Very soon we might be seeing a day where the automated stenographer would get promoted and start taking an active part in organizing the meetings and presentations.
Thanks to the evolution of speech recognition technology, there’s an easier and faster way of converting speech into text—such as for generating transcripts of audio content. Machines can perform this time-consuming tedious job accurately and quickly.
So who needs speech to text functionality?
Audio sections of libraries generate transcripts of faculty lectures and guest talks for students’ reference. Organizations like TED generate English transcripts of TED Talks, which make it easier for people to translate the talks into other languages for a wider reach. Broadcasters generate closed captioning, also called subtitling for television programming, which lets the deaf and those who are hard of hearing follow the news and shows.
The beauty of transcripts is that they increase the potential audience of single language video content. Generating multilingual video subtitles and adding the option to display one of these per the consumer’s preference makes the content consumable by those who do not speak the language of the original audio stream. Incidentally, research shows that subtitling can ensure that more people watch a video to its completion.
Transcripts also make content consumable in mute mode. For instance, Oracle uses speech to text technology to generate closed captioning for proceedings of its enthusiastically followed technology conferences: Oracle Partner KickOff, JavaOne, and Oracle OpenWorld. It acknowledges that subtitling gives “a global audience” access to the conference proceedings as well as “viewers tuning in at their desks requiring muted volume.”
Another plus point of transcripts is they enhance the value of content by making it searchable for certain words and phrases. Search engines like Google, Bing and even a company’s internal search engine can fully index transcripts, and thus make previously inaccessible audio content accessible. For instance, a civil engineering student might want to search faculty lectures for the keywords “green building” or “glass house,” which is possible if the lectures have been transcribed.
Building speech to text functionality into apps could allow individual consumers to generate transcripts of live audio streams for their own consumption—such as helping attendees of multilingual conferences or participants of multilingual conference calls understand what’s going on.
GoVivace’s Speech to Text offers large vocabulary speech recognition technology that is especially well suited to generating accurate subtitling or audio transcripts and thus speeding up what was once a labor-intensive job. Accurately converting speech into text requires a speech engine that has a large and preferably, evolving vocabulary, so that it can understand practically any word that is fed to it. The GoVivace speech recognition engine fits the bill.
Spare yourself from the listen-type-rewind cycle and raise the bar on subtitles in the process.
511 services offer information ranging from today’s weather to “how soon will the L7 bus be here”. Today millions of travelers rely on 511 every month, to retrieve traffic updates, transit information and alternative solutions to their transportation needs. 511 is an integral part of what is a larger effort to create “Smart Cities” throughout the world. Currently, IBI Group is instrumental in building “Smart Cities” in 11 countries. IBI is considered the industry leader in designing, organizing and producing 511 systems throughout the United States, most notably in the Greater Los Angeles area, (LA, Orange, and Ventura counties), as well as a 511 systems for the states of Florida, New York, Massachusetts, and Alaska.
The beauty of IBI Group’s 511 interactive voice response(IVR) system is the ability to tailor each system for a particular region’s needs. Customization allows for a client to offer a handful of options in a rural area while a more advanced, multi-layer system is imperative for metropolitan areas. It is to be expected that users of this system will not be calling from quiet locations but rather noisy environments such as train stations, bus stops or at the side of the road. Therefore, the speech recognition technology utilized must include the first-rate algorithm to filter out the noise and 21st-century technology to interpret utterances into recognizable text. IBI Group has repeatedly chosen GoVivace.
Utilizing proprietary neural networks and deep learning techniques, GoVivace has been able to interpret the noisiest utterances, guiding the Voice User Interface (VUI) to find the appropriate response, thereby completing a caller’s request expeditiously and accurately. The GoVivace’s speech engine can handle a large vocabulary, or a constrained topic specific language model, which provides for a higher rate of accuracy when the application permits.
Larry Baldwin, Manager of Voice Services at IBI Group, states “I chose, and will continue to choose, GoVivace for three main reasons: accuracy, service, and price.
I’ve found that the GoVivace team has produced a very high-quality product and their team is very supportive”.
The path from choosing a speech technology vendor to deploying a 511 solution was a long journey. Data sources that form the catalog and menu of the system came from disparate sources. All that data needed to be normalized to an easily readable format. As is the nature of large cities, the names of the streets are not regular English words and the vocabulary sizes can be large. The resulting grammars were gigantic, and efficient optimization was needed to make everything work smoothly. IBI and GoVivace worked together to sort through all those issues.
GoVivace is the go-to speech provider that offers the combination of grammar based and large vocabulary speech recognition (say anything) with highly customized language models. This ensures optimum accuracy in noisy environments for highly vertical subjects such as finance, travel, and medical. GoVivace can also fit in the smallest domain, from silicon to IoT devices, while highly scalable on cloud services like AWS©or Azure©.
In the internet age, loss of personal credentials through security breaches (such as the Target Data Breach in 2013) has become a common occurrence. This places individuals and financial institutions at an elevated risk of fraudulent activity.
However, some of this activity may be stopped in its tracks by employing additional voice biometrics. Consider that case when a user calls the wire transfer line to request a wire. The user presents some “secret” information and then is able to complete the transaction. However, under this setup, the secret information could easily be stolen and then used against the account owner.
However, if the bank deploys a voice biometric system, not only will the credentials of the user be stored in the system’s database, by also a small biometric key of the user that will be less than a kilobyte in size, shall contain the voiceprint of the user. Every time the user calls for a transaction, his voice shall be analyzed and a match will be created to tell if it is the same user. If an imposter makes the call, the system will clearly warn the agent to increase the security level and use stricter authentication methods.
There are several features of modern voice biometric systems that make such a system not a dream of the future, but a real implementation that is in the process of happening. First, the voiceprints are small. It takes less than a kilobyte to store the voiceprint. It can be added to the same database that stores the passwords and other identifying information of the user, without much bloat in database size.
Second, the process of computing the match is extremely fast. A ten second of conversation at the start of the call may take just a fraction of a second to compare with the stored voiceprint and tell if it is a likely match.
Third, there used to be a time when the same speaker will be easily rejected if they use a different handset to make the call or if the telephone channel is facing issues. However recent advances in voice biometrics have significantly reduced the impact of such factors while maintaining the inter-speaker variability.
Financial institutions are already required to record conversations with their clients and store them for a certain period of time. If an institution determines to enhance their level of security through voice biometrics system, these stored recordings could be used to enroll the voiceprints of all of their existing users. There will be no change to the user experience, though the call center agents will have an additional authentication metric available to them that allows the institution an additional layer of security without compromising convenience.
Texting thumb is every smartphone user’s nightmare. All because mobile apps have invaded every facet of modern lives—socializing, shopping, banking, and so on. And, they promise to get still more versatile. Voice technology is predicted to power the next wave of mobile apps.
Consider for example a grocery store shopping mobile app. What every smartphone owner could use is a voice-enabled version, which would accept spoken input instead of compelling the user to fumble on a tiny keyboard. And the app would actually understand and respond to what has been said. Using an online grocery store’s app, they could just rattle off their shopping list and let the app search for the items and fill the cart. Then they could command the app to proceed to checkout, and voila! Without a single keystroke, the chore is done (leave aside waiting for the home delivery). No fumbling, no worn out thumbs. Nothing beats hands-free voice input—wouldn’t you agree?
Can an app really be so user-friendly?
Welcome to the talky online world populated by the likes of Google voice and Siri, Apple’s personal assistant app. Technology to voice-enable mobile apps is already a reality.
GoVivace Inc. of McLean, VA has designed an advanced Automatic Speech Recognition engine that is specifically suited for voice-enabled mobile apps. Why’s that?
The key to building reliable and robust voice-enabled mobile apps is to construct a comprehensive application grammar and vocabulary, technical jargon for a set of pre-specified possibilities that the app will look up to understand the speech input. The more inclusive the grammar, the better the app will understand and seem intelligent to the user!
Say the user asks the voice-enabled mobile app of an online grocery store to add two packets of Oreo’s stuffed chocolate chip cookies to the shopping cart. A number of things happen behind the scenes. Just like Siri or any other voice-enabled mobile app, the audio stream representing the spoken input is compressed and sent to a waiting farm of servers. Those servers have also notified the context in which the input was spoken. Putting together the context and the input, the servers quickly change their language model to suit the situation and then convert the audio into text. The servers recognize that “two packets” is the quantity and “Oreo’s stuffed chocolate cookies” is the name of the item.
Essentially, the item is looked up in the apps grammar and vocabulary representing the hundreds and even thousands of possible inputs the servers may have to process, and finally, the cart is updated. It involves a lot of steps but everything happens so quickly that the app user doesn’t notice the number of steps involved, and happily goes on shopping.
The performance of voice-enabled mobile apps also depends on the quality of the speech recognition engine. Ideally, the engine must be capable of understanding natural language and adapting to variations in voice quality and spoken content. At the same time, it should be easy to use and integrate.
GoVivace’s Automatic Speech Recognition uses both grammars and a statistical language model to understand natural language, which helps build highly precise voice-enabled mobile apps.
Big Data, higher compute capacities and GPUs and deep learning has had a transformative effect on the speech technologies of today. The transformation in technology has lead to accuracies that only rival humans. Obviously, the industry has moved to embrace the new technology and the effect is visible in contact centers across the globe.
For instance, whenever we make a call to an enterprise or even to customer service, we must have noticed that a physical human being never answers our call. Instead, we receive an automated voice which records answers and gives instructions to press buttons and guides us through a built-in menu.
In recent years, all this has been possible only due to the emergence of a new technological breakthrough in the field of Machine Learning and Signal Processing popularly known as Automatic Speech Recognition.
Automatic Speech Recognition (ASR) and related subjects are continuously evolving and have become inseparable elements of Human-Computer Interaction (HCI). With the conjunction of emerging concepts like Big Data and the Internet of Things (IoT), learning based techniques are becoming the hot topic for research communities in the field ASR. Though there is plenty of commercial voice recognizers available for certain applications like dictation and transcription, due to many issues in ASR like recognition in noisy environments, multilingual recognition, and multi-modal recognition a near to idea ASR engine is yet to be addressed effectively.
The various machine learning techniques like Artificial Neural Networks(ANN), support vector machines and Gaussian Mixture Models(GMM) along with Hidden Markov Models(HMM) employed in ASR systems.
Automatic Speech Recognition engine is capable of recognizing spoken words and converting the recorded voice into text and it also supports the real-time audio transcription. The ASR engine can support various languages and accents that can be localized to any language. The ASR engine is capable of actioning voice commands given to electronic devices such as computers, PDAs or telephones through microphones. Though the most recent version of ASR technology is based on Natural Language Processing(NLP) and this technical leap in voice recognition technology comes closest to allowing real conversation between a person and machine.
Though it can’t be deduced that it is the final step. Rather it’s still a long way to cover before reaching an apex of development, we’re already witnessing some remarkable results in the form of intelligent smartphone interfaces like the Siri, Alexa or Google assistant program.
The engine compares the spoken input with a number of pre-specified possibilities. The various pre-specified possibilities constitute the application’s grammar, which drives the interface between the dialogue-speaker and the back-end processing. The voice recognition solution supports both simple grammar and very large grammars for complex tasks such as dates, complex commands, and yellow pages styled complex directory lookups. Both types of grammars can be stored on the server after compilation, to ensure fast processing for future utterances. It makes use of statistical and neural language model to understand natural language. That’s all because of voice recognition engines that operate using algorithms through acoustic and linguistic modeling.
Acoustic modeling signifies the relationship between the linguistic units of speech and audio signals whereas Language modeling matches the sounds with word sequences to distinguish between words that sound similar. This software can be used in offices and businesses, which can enable users to speak to their computers and have their words converted into text via voice recognition. You can access function commands like setting a meeting or conference, opening files, making a call to the client and much more.
So, what is keeping speech recognition from becoming dominant?
Truly, it’s the challenges imposed by this technology of which accent and voice are biggest and these aren’t accessible by word recognition platforms. And simply recognizing voice is not enough — the software must also recognize new words and proper nouns. But players like IBM, Google and GoVivace Inc. of McLean, Virginia are leading today’s era of speech with their voice recognition technologies and solutions in B2B markets which in turn is positively making common man’s life easy and more productive in a scalable manner.
As Larry Baldwin, Manager of Voice Services at IBI Group, states “I chose and will continue to choose, GoVivace for three main reasons: accuracy, service, and price. I’ve found that the GoVivace team has produced a very high-quality product and their team is very supportive.”
The IBI group deployed 511 systems throughout the United States, most notably in the Greater Los Angeles area, (LA, Orange, and Ventura counties), as well as for the state of Florida, New York, Massachusetts, and Alaska. 511 services offer a plethora of information, ranging from today’s weather to “how soon will the L7 bus be here”. Today millions of commuters use 511 services daily, whether to check traffic updates, transit information or alternative solutions to their transportation needs. Although the users of this system will not be calling from quiet locations, rather from noisy environments such as train stations, bus stops or at the side of the road.
Therefore, the speech recognition technology utilized must include the first-rate algorithm to filter out the noise and 21st-century technology to interpret utterances into recognizable text that is audio transcription by speech to text engine.
Utilizing proprietary neural networks and deep learning techniques, it has been able to interpret the noisiest utterances, guiding the Voice User Interface (VUI) to find the appropriate response, thereby completing a caller’s request expeditiously and accurately. The GoVivace’s automatic speech recognition engine with an accuracy rate of 87 percent can handle a large vocabulary, or a constrained topic specific language model, which provides for a higher rate of accuracy when the application permits.
Regardless of how much advancements there have been till now or going on ASR technology, there’s still a long way to go as speech recognition comes with numerous challenges such as human-like understanding of voice as humans have their own knowledge base which is resulted from reading, experiments, experiences, examination, situation, interaction and communication. They may hear more than the speaker speaks to them. While speaking speakers have its own language model of the native language. Humans may understand and interpret the words or sequence of words, they never heard before, but that’s not the case with ASR now as the models have to be trained according to certain requirements.
But, as mankind has challenged technology over since the last two centuries, it will surely be going to conquer these challenges and will define the whole new definition of speech and GoVivace Inc. is contributing to the speech technology since a decade.