Robotic Process Automation

What Is Robotic Process Automation (RPA)?

Robotic process automation (RPA) occurs when basic tasks are automated through software or hardware systems that function across a variety of applications, just as human workers do. This can greatly reduce labor costs and increase efficiency by speeding things up and greatly minimizing human error. The software or robot can be taught a workflow with multiple steps and applications, such as taking received forms, sending a receipt message, checking the form for completeness, filing the form in a folder, and updating a spreadsheet with the name of the form, the date filed, and so on. RPA software is designed to reduce the burden for employees of completing repetitive, simple tasks.

1) Robotic process automation (RPA) refers to software that can be easily programmed to do basic, repetitive tasks across applications.

2) RPA creates and deploys a software robot with the ability to launch and operate other software.

3) Designed primarily for office-type functions, RPA works like a digital assistant, doing routine onerous tasks that would otherwise eat up employees' time.

4) RPA today is found across a range of industries and applications.

5) RPA without human oversight, however, can lead to problems, as was the case with mortgage "robot-signers."

Robotic process automation (RPA) is designed to help primarily with office-type functions that often require the ability to do several types of tasks in a specific order. It creates and deploys a software robot with the ability to launch and operate other software. In a sense, the basic concept is similar to traditional manufacturing automation, which focuses on taking one portion of a workflow—or even just one task—and creating a robot to specialize in doing it.

Benefits of Robotic Process Automation (RBA)

The software used in process automation is programmed to do the tasks in a particular workflow by the employees with minimal assistance from human workers. The software doesn’t learn on its own or seek to tweak out new efficiencies or new insights like big data analysis or enterprise resource management (ERM) software. Instead, RPA works like a digital assistant for employees by clearing the onerous, simple tasks that eat up part of every office worker’s day.

As such, RPA is a simpler product than an artificial intelligence-driven system or enterprise software that seeks to bring all data inside the platform. This also makes it a relatively cheaper product than AI or ERM software. This simplicity and relative cheapness can make RPA a more attractive solution for many companies, particularly if the company has legacy systems. Robotic process automation is designed to be compatible with most legacy applications, making it easier to implement compared to other enterprise automation solutions.

RBA Be Applied

RBA is quite common in the financial services industry. With increasing compliance and regulatory filing requirements, the finance industry—banks, insurers, and investment management companies—has been an early adopter of RPA. Many onerous back-office functions, such as ensuring an up-to-date know your client (KYC) form is filed or a recent credit check is included on a loan application, are ideal for RPA. Removing this burden from employees allows them to focus on high-return tasks. More importantly, the software can clear these basic filing and data manipulation functions faster than humans, reducing the overall processing time.Of course, RPA is not just limited to finance. Any industry that deals in data and filing can benefit from robotic process automation. When software can reduce costs and increase efficiency without requiring an onerous and complex implementation, it will find eager users and useful applications in almost any sector. Indeed, RPA has also been found useful in the following fields:

• Customer service and CRM
• Accounting
• Healthcare
• Human resources
• Supply chain management

RPA does, however, have its drawbacks. These systems can be expensive to customize and deploy, and may not be suitable for more complex tasks that require some degree of human judgment or creativity.RPA systems, when unchecked, can also go awry. One example is the case of so-called "robo-signers" used in the mortgage industry. These systems rubber-stamped foreclosure documents on homeowners automatically, even when the foreclosure was questionable or avoidable. Moreover, this practice failed to meet government regulations for oversight of the foreclosure process in the mid-2010s, leading to a scandal following the housing market bubble of the 2008-09 financial crisis. Following the public exposure of Robo-signers, foreclosure documents had to be manually reexamined, and the companies involved faced disciplinary action.

Machine Learning

What Is Machine Learning

The robot-depicted world of our not-so-distant future relies heavily on our ability to deploy artificial intelligence (AI) successfully. However, transforming machines into thinking devices is not as easy as it may seem. Strong AI can only be achieved with machine learning (ML) to help machines understand as humans do.

Machine Learning Work

Similar to how the human brain gains knowledge and understanding, machine learning relies on input, such as training data or knowledge graphs, to understand entities, domains, and the connections between them. With entities defined, deep learning can begin. Machine learning begins with observations or data, such as examples, direct experience, or instruction. It looks for patterns in data so it can later make inferences based on the examples provided. The primary aim of ML is to allow computers to learn autonomously without human intervention or assistance and adjust actions accordingly.

Machine Learning Important

Machine learning as a concept has been around for quite some time. The term “machine learning” was coined by Arthur Samuel, a computer scientist at IBM and a pioneer in AI and computer gaming. Samuel designed a computer program for playing checkers. The more the program played, the more it learned from experience, using algorithms to make predictions. As a discipline, machine learning explores the analysis and construction of algorithms that can learn from and make predictions on data.ML has proven valuable because it can solve problems at a speed and scale that cannot be duplicated by the human mind alone. With massive amounts of computational ability behind a single task or multiple specific tasks, machines can be trained to identify patterns in and relationships between input data and automate routine processes.

Data Is Key:

The algorithms that drive machine learning are critical to success. ML algorithms build a mathematical model based on sample data, known as “training data,” to make predictions or decisions without being explicitly programmed to do so. This can reveal trends within data that information businesses can use to improve decision-making, optimize efficiency and capture actionable data at scale.

AI Is the Goal: 

ML provides the foundation for AI systems that automate processes and solve data-based business problems autonomously. It enables companies to replace or augment certain human capabilities. Common machine learning applications you may find in the real world include chatbots, self-driving cars, and speech recognition.

• Data security: 
• Machine learning models can identify data security vulnerabilities before they can turn into breaches. By looking at past experiences, machine learning models can predict future high-risk activities so the risk can be proactively mitigated.
• Finance: 
• Banks, trading brokerages, and fintech firms use machine learning algorithms to automate trading and provide financial advisory services to investors. Bank of America is using a chatbot, Erica, to automate customer support.
• Healthcare: 
• ML is used to analyze massive healthcare data sets to accelerate the discovery of treatments and cures, improve patient outcomes, and automate routine processes to prevent human error
• Fraud detection: 
• AI is being used in the financial and banking sector to autonomously analyze large numbers of transactions to uncover fraudulent activity in real-time. Technology services firm Capgemini claims that fraud detection systems using machine learning and analytics
• Retail: 
• AI researchers and developers are using ML algorithms to develop AI recommendation engines that offer relevant product suggestions based on buyers’ past choices, as well as historical, geographic, and demographic data.

Machine learning offers clear benefits for AI technologies. But which machine learning approach is right for your organization? There are many ML training methods to choose from including:

• supervised learning
• unsupervised learning
• semi-supervised learning

Supervised Learning: 

Supervised machine learning algorithms apply what has been learned in the past to new data using labeled examples to predict future events. By analyzing a known training dataset, the learning algorithm produces an inferred function to predict output values. The system can provide targets for any new input after sufficient training. It can also compare its output with the correct, intended output to find errors and modify the model accordingly.

Unsupervised Learning: 

Unsupervised machine learning algorithms are used when the information used to train is neither classified nor labeled. Unsupervised learning studies how systems can infer a function to describe a hidden structure from unlabeled data. At no point does the system know the correct output with certainty. Instead, it draws inferences from datasets as to what the output should be

Reinforcement Learning: 

Reinforcement machine learning algorithms are a learning method that interacts with its environment by producing actions and discovering errors or rewards. The most relevant characteristics of reinforcement learning are trial and error search and delayed reward. This method allows machines and software agents to automatically determine the ideal behavior within a specific context to maximize its performance. Simple reward feedback — known as the reinforcement signal — is required for the agent to learn which action is best.

Biometrics Technologies

Biometrics is the measurement and statistical analysis of people's unique physical and behavioral characteristics. The technology is mainly used for identification and access control or for identifying individuals who are under surveillance.

How do biometrics work?

Authentication by biometric verification is becoming increasingly common in corporate and public security systems, consumer electronics and point-of-sale applications. In addition to security, the driving force behind biometric verification has been convenience, as there are no passwords to remember or security tokens to carry. Some biometric methods, such as measuring a person's gait, can operate with no direct contact with the person being authenticated.

Components of biometric devices include the following:

1) a reader or scanning device to record the biometric factor being authenticated

2) software to convert the scanned biometric data into a standardized digital format and to compare match points of the observed data with stored data; and

3) a database to securely store biometric data for comparison.

Biometric data may be held in a centralized database, although modern biometric implementations often depend instead on gathering biometric data locally and then cryptographically hashing  it so that authentication or identification can be accomplished without direct access to the biometric data itself.

Types of biometrics

The two main types of biometric identifiers are either physiological characteristics or behavioral characteristics.Physiological identifiers relate to the composition of the user being authenticated and include the following:

1) facial recognition

2) fingerprints

3) finger geometry

4) iris recognition

5) vein recognition

6) retina scanning

7) voice recognition

8) DNA (deoxyribonucleic acid) matching

9) digital signatures

Behavioral identifiers include the unique ways in which individuals act, including recognition of typing patterns, mouse and finger movements, website and social media engagement patterns, walking gait and other gestures. Some of these behavioral identifiers can be used to provide continuous authentication instead of a single one-off authentication check. While it remains a newer method with lower reliability ratings, it has the potential to grow alongside other improvements in biometric technology.

Biometric data can be used to access information on a device like a smartphone, but there are also other ways biometrics can be used. For example, biometric information can be held on a smart card, where a recognition system will read an individual's biometric information, while comparing that against the biometric information on the smart card.

Advantages and disadvantages of biometrics

The use of biometrics has plenty of advantages and disadvantages regarding its use, security and other related functions. Biometrics are beneficial for the following reasons:

• hard to fake or steal, unlike passwords;
• easy and convenient to use;
• generally, the same over the course of a user's life;
• nontransferable; and
• efficient because templates take up less storage.

      Disadvantages, however, include the following:

• It is costly to get a biometric system up and running.
• If the system fails to capture all of the biometric data, it can lead to failure in identifying a user.
• Databases holding biometric data can still be hacked.
• Errors such as false rejects and false accepts can still happen.
• If a user gets injured, then a biometric authentication system may not work -- for example, if a user burns their hand, then a fingerprint scanner may not be able to identify them.
• Law enforcement.It is used in systems for criminal IDs, such as fingerprint or palm print authentication systems.
• United States Department of Homeland Security. It is used in Border Patrol branches for numerous detection, vetting and credentialing processes -- for example, with systems for electronic passports, which store fingerprint data, or in facial recognition systems.
• Healthcare. It is used in systems such as national identity cards for ID and health insurance programs, which may use fingerprints for identification.
• Airport security. This field sometimes uses biometrics such as iris recognition.

Security and privacy issues of biometrics

Biometric identifiers depend on the uniqueness of the factor being considered. For example, fingerprints are generally considered to be highly unique to each person. Fingerprint recognition, especially as implemented in Apple's Touch ID for previous iPhones, was the first widely used mass-market application of a biometric authentication factor.Other biometric factors include retina, iris recognition, vein and voice scans. However, they have not been adopted widely so far, in some part, because there is less confidence in the uniqueness of the identifiers or because the factors are easier to spoof and use for malicious reasons, like identity theft.

Stability of the biometric factor can also be important to acceptance of the factor. Fingerprints do not change over a lifetime, while facial appearance can change drastically with age, illness or other factors.The most significant privacy issue of using biometrics is that physical attributes, like fingerprints and retinal blood vessel patterns, are generally static and cannot be modified. This is distinct from nonbiometric factors, like passwords (something one knows) and tokens (something one has), which can be replaced if they are breached or otherwise compromised. A demonstration of this difficulty was the over 20 million individuals whose fingerprints were compromised in the 2014 U.S.office of personnel mangagement data breach.

The increasing ubiquity of high-quality cameras, microphones and fingerprint readers in many of today's mobile devices means biometrics will continue to become a more common method for authenticating users, particularly as Fast ID Online has specified new standards for authentication with biometrics that support two-factor authentication with biometric factors.While the quality of biometric readers continues to improve, they can still produce false negatives, when an authorized user is not recognized or authenticated, and false positives, when an unauthorized user is recognized and authenticated.

Biometrics secure

While high-quality cameras and other sensors help enable the use of biometrics, they can also enable attackers. Because people do not shield their faces, ears, hands, voice or gait, attacks are possible simply by capturing biometric data from people without their consent or knowledge.An early attack on fingerprint biometric authentication was called the gummy bear hack, and it dates back to 2002 when Japanese researchers, using a gelatin-based confection, showed that an attacker could lift a latent fingerprint from a glossy surface. The capacitance of gelatin is similar to that of a human finger, so fingerprint scanners designed to detect capacitance would be fooled by the gelatin transfer.Determined attackers can also defeat other biometric factors. In 2015, Jan Krissler, also known as Starbug, a Chaos Computer Club biometric researcher, demonstrated a method for extracting enough data from a high-resolution photograph to defeat iris scanning authentication. In 2017, Krissler reported defeating the iris scanner authentication scheme used by the Samsung Galaxy S8 smartphone. Krissler had previously recreated a user's thumbprint from a high-resolution image to demonstrate that Apple's Touch ID fingerprinting authentication scheme was also vulnerable.

Decision Management

Management Decision Making

Management decision-making is a critical part of the management planning function. Understanding the unique nature of managerial decisions requires understanding the types of decisions and the context for making those decisions.

Types or Categories of Management Decisions

Decision-making can be defined as selecting between alternative courses of action. Management decision-making concerns the choices faced by managers within their duties in the organization. Making decisions is an important aspect of planning. Decision-making can also be classified into three categories based on the level at which they occur.

Strategic Decisions

These decisions establish the strategies and objectives of the organization. These types of decisions generally occur at the highest levels of organizational management.

Tactical Decisions

Tactical decisions concern the tactics used to accomplish the organizational objectives. Tactical decisions are primarily made by middle and front-line managers.

Operational Decisions

Operational decisions concern the methods for carrying out the organization's delivery of value to customers. Operational decisions are primarily made by middle and front-line managers.

Decisions can be categorized based on the capacity of those making the decision.

Organizational Decisions 

An organizational decision is one that relates or affects the organization. It is generally made by a manager or employee within their official capacity. These decisions are often delegated to others.

Personal Decisions

Personal decisions are those primarily affecting the individual - though the decision may ultimately have an effect on the organization as a result of its effect on the individual. These types of decisions are not made within a professional capacity. These decisions are generally not delegated to others.

Areas of Decision Management

The goal of decision management is to enhance business operations intelligence by ensuring quick, consistent, and accurate fact-based decisions. The quality of structured operational decisions, no matter how complex, should be constantly improving. There are five areas that affect decision management:

Data and analytics: Data is accessed and processed with the help of descriptive, diagnostic, and predictive techniques. You need strong data quality as a basis for accurate decision-making, and the outcomes of those decisions affect the data as well.

Business Management Process: Managing human tasks and the sequence of business process automation and task management. The information from staff helps to make better decisions, and their roles are enhanced as a result.

Operations research: Optimizing and managing various goals based on standards and priorities that can be modeled. Decision management analyzes operations and suggests improvements that can be made.

Business rules management: Automating business rules and managing them based on inputs provided by subject matter experts.

Robotics: Using software to imitate human behavior in the automation of actions and related interactions with software systems.

Decision management results in efficiency and productivity, two critical factors for successful business operations. As a concept, decision management can be used in a wide number of industries, functions, and areas of business. There are so many businesses that make scores of operational decisions on a daily basis. The quality of these decisions has a direct impact on the effectiveness of the company. All decisions are impacted by data, regulations, market dynamics, and decision management—and therefore becomes a necessity.

Benefits of Decision Management

Better Utilization of Time

Regardless of the model of the decision management support system, research shows that it reduces the decision time cycle. Employee productivity is the immediate benefit from the time saved.

Better Efficacy

The effectiveness of decisions made with decision management is still debated because the quality of these decisions is hard to measure. Research has largely taken up the approach of examining soft measures like a perceived decision quality instead of objective measures. Those who advocate the creation of data warehouses are of the strong opinion that better and larger-scale analyses can definitely enhance decision-making.

Better Interpersonal Communication

Decision management systems open the door for better communication and collaboration among all decision-makers. Set rules ensure that all decision-makers are on a single platform, sharing facts and any assumptions made. Data-driven rule sets analyze and provide decision-makers with the best version of the possible outcome, encouraging fact-based decision-making. Better access to data always enhances the quality and clarity of decisions.

Cost Reduction

An outcome of good decision management rule sets is saving costs in labor (which comes from good decision-making, lowered infrastructure, and technological costs).

Better Learnings

In the long term, a by-product of decision management is that it encourages learning. There is more openness to new concepts, and a fact-based understanding of businesses, and the overall decision-making environment. Decision management can also come in handy to train new employees—an advantage yet explored in full.

Increased Organizational Control

With decision-making rule sets, a lot of transactional data is made available for constant performance checks and ad hoc inquiries by business heads. This gives management a better look at how business operations work. Managers find this to be a useful aspect of decision-making. There is a financial benefit to highly-detailed data, and this gradually becomes evident.

Disadvantages of Decision Management

As with any system, decision management systems can have a few disadvantages.

Information Overload

Considering the amount of data that goes through the system (and the fact that a problem is analyzed from multiple aspects), there are chances of information overload. With too many variables available on hand, the decision maker may be faced with a dilemma. Streamlined rule sets can help.

Over-Dependence

When decision-making is completely computer-based, it can lead to over-dependence. While it does free up man hours for better use of skills, it also increases dependency on computer-based decision-making. Individuals can be less inclined to think independently and come to rely on computers to think for them.

Subjectivity

One of the important aspects of decision-making is the number of alternatives that are offered based on objectivity. Subjectivity then tends to take a backseat, and this can affect decision-making and impact businesses. Things that cannot be measured cannot be factored in.

Overemphasis on Decision Making

Not all issues an organization is faced with need the power of decision management. An emphasis has to be placed on utilizing decision-making capabilities for relevant issues.

Types of Decision Support Systems for Decision Making

Decision support systems are classified into two types

Model-Based Decision Support Systems: These stand independent of any corporate information system. They work on the basis of strong theory or models and come with an excellent interface for easy interactivity

Data-Based Decision Support Systems: These set-ups collect large amounts of data from a variety of sources, store it in warehouses, and analyze it. The warehouse stores historical data and also comes with some reporting and query tools.

In data-based decision support systems there are two main techniques that are employed:

Online Analytical Processing (OLAP): Based on queries, this provides quick answers to some complex business needs. Managers and analysts can actively interact and examine data from multiple viewpoints.

Data Mining: By finding patterns and rules in existing data, useful decision-making information can be extracted to help in trend and consumer behavior patterns.

Virtual Agents

A virtual agent (sometimes called an intelligent virtual agent (IVA), virtual rep, or chatbot) is a software program that uses scripted rules and, increasingly, Artificial Intelligence applications to provide automated service or guidance to humans.

Virtual agents are most commonly used by organizations in their customer service functions to answer routine customer queries, fulfill standard requests, and/or handle simple problems. For example, virtual agents are often used for initial customer interactions with call centers or click-to-chat features on websites. Virtual agents are also used in some organizations to handle employee-driven needs. For example, virtual agents are commonly deployed within the IT function to provide help desk-type services, such as employee requests for resetting computer passwords. They can also be used in organizations to guide employees through work tasks or processes. In this way, a virtual agent is akin to a digital assistant, an application program that understands natural language voice commands and is also deployed to fulfill people's needs or help them complete tasks.

Technology research and advisory firm Gartner predicted that 25% of customer services and support operations will use virtual assistants across their engagement channels in 2020, up from less than 2% in 2017. In addition, 25% of digital workers will use virtual assistants in their tasks on a daily basis by 2021, compared with less than 2% in 2019, according to Gartner.

Virtual agent vs. virtual assistant

The terms virtual agent and virtual assistant are often used interchangeably with each other, as well as with the term "chatbot." Although all three are types of computerized aid offered to serve people in various capacities, there are some subtle (although not definitive or universally accepted) distinctions between each of the terms. Virtual agent and virtual assistant are more closely aligned terms and, thus, more likely to be used interchangeably. However, many associates the term virtual assistant with Apple's  Siri, Amazon's Alexa, and Google Assistant -- all platforms that draw on the internet and other technologies to perform internet searches and digital tasks, such as updating calendars or checking weather forecasts in response to a user's request. The term virtual agent, on the other hand, is more commonly associated with organizational use, where agents are put to work assisting customers or employees. A chatbot is a specific type of virtual agent -- a conversational agent -- with capabilities to "chat" either via email or messaging or voice. However, the term "chatbot" does not encompass the wider array of virtual agent capabilities that might also include visual representations such as a hologram, as well as other additional characteristics beyond verbal communication. The term "virtual agent" can also refer to a human agent who works remotely from his or her employer's location to serve customers.

How virtual agents work

Virtual agent technologies initially emerged in the first decade of the 2000s. At the most basic level, virtual agent technologies work on a preprogrammed scripted model. Organizations could create virtual agents that were scripted to respond in specific ways to specific human requests. Organizations generally identified the particular workflows that would be handled by the virtual agents, mapping out what a virtual agent should do based on each specific request or inquiry made by a person. Organizations then created the scripts to have the agent respond as needed to each request, which the agent could identify by predetermined keywords that had been programmed into the platform. In other words, the virtual agent would identify the keywords and respond with the scripted response that in its computerized analysis best matches the keywords.

As such, these virtual agents could handle routine tasks where an inquiry or request could be met with a predictable response. Organizations programmed their virtual agents to turn over the customer interaction to human agents when requests hit a certain point in the workflow or when the inquiries digressed from the script. In the second decade of the 2000s, particularly toward the latter half, virtual agent platforms incorporated machine learning, natural language processing, and artificial intelligence to create intelligent virtual agents that could handle more types of queries, as well as less predictable inquiries, requests, and workflows. These intelligent virtual agent platforms can also connect with back-end systems, thereby providing more personalized responses to the customers or employees who are interacting with the agent systems. Moreover, the AI capabilities built into these platforms enable these agents to "learn," so they can become more efficient and effective as they work, and they can also develop the capacity to handle a wider range of tasks.

Virtual agent capabilities

As virtual agent software has improved in the second half of the 2010s with advances in AI and cognitive computing programs, virtual agents have moved far beyond interactive voice response IVR  systems. In fact, technological advances have enabled virtual agents to understand customer intent and can provide personalized answers to customer questions in a human-like manner. However, virtual agents still typically communicate with customers via email or live chat on corporate websites. Companies may also use an avatar to provide a visual representation of the virtual agent. Additionally, most companies as of 2020 still use virtual agents to handle highly repeatable tasks. For complicated tasks, live customer service agents are required. In the world of customer relationship management CRM  software, virtual agents are used to providing 24/7 customer service, including answering questions on accounts, helping with a password, providing recommendations, or following up on sales and marketing leads via email correspondence. For example, a virtual sales agent can be used to email potential customers to request a meeting with a live sales agent. When a customer agrees to a meeting, the virtual agent can obtain a phone number and collect the information a sales rep might need to conduct a live conversation. This is enormously useful for sales and marketing teams, as they typically only focus on leads deemed "high quality." With a virtual agent, all leads can be followed up on, which could result in higher sales. In addition, virtual agents cost significantly less than human employees.

How to use a virtual agent

Companies interested in adopting virtual agent software through a cloud service provider or software vendor must invest time and resources into "training" the virtual agent. This initial setup period may take months to complete, depending on the level of confidence the company desires. Virtual agents are based on machine learning technology, which improves over time as the system ingests more data and "learns" through continued use. Virtual agents can only provide information that has been "fed" to the AI system, and if the system contains bad data, customers will receive false information. This makes the setup phase critical. The initial time investment is worthwhile when it results in reduced call volume and frees up live agents to focus on complex customer service tasks, while simultaneously providing a  good customer experience.

Speech Recognition & History of Voice Recognition Technology

What is Speech Recognition

Speech recognition software is a computer program that’s trained to take the input of human speech, interpret it, and transcribe it into text.

How Does It Work

Speech recognition software works by breaking down the audio of a speech recording into individual sounds, analyzing each sound, using algorithms to find the most probable word fit in that language, and transcribing those sounds into text. Speech recognition software uses natural language processing (NLP) and deep learning neural networks. “NLP is a way for computers to analyze, understand, and derive meaning from human language in a smart and useful way,” according to the algorithms blog. This means that the software breaks the speech down into bits it can interpret, converts it into a digital format, and analyzes the pieces of content. From there, the software makes determinations based on programming and speech patterns, making hypotheses about what the user is actually saying. After determining what the users most likely said, the software transcribes the conversation into text. This all sounds simple enough, but the advances in technology mean these multiple, intricate processes are happening at lightning speed. Machines can transcribe human speech more accurately, correctly, and quickly than humans can.

Speech Recognition & AI Software

Voice recognition and transcription technology have come a long way since their first inception. We now use voice recognition technology in our everyday lives with voice search on the rise, more people are using assistants like Google Home, Siri, and Amazon Alexa.

History of Voice Recognition Technology

Programmers and engineers have made great leaps in the science of voice recognition over the past decade, so you’d be forgiven for thinking that this technology is a relatively new development. Much of the reporting and scholarship around voice recognition tech only focuses on the post-2011 Age of Siri, following the release of Apple’s now-ubiquitous personal assistant.

But there’s a rich secret history to voice recognition tech that stretches back to the mid-20th-century, to those early days when rudimentary computers needed to fill an entire warehouse with vacuum tubes and diodes just to crunch a simple equation. And this history not only reveals some interesting trivia about the technology we know and love today, but it also points the way toward potential future breakthroughs in the field. Let’s explore the untold story of voice recognition technology, and see how much progress has been made over the years (and how much has stayed the same).

AUDREY and the Shoebox

In the early 20th century, the U.S. research firm Bell Laboratories (named after founder Alexander Graham Bell, the inventor of the telephone) racked up a string of impressive technological advances: The invention of radio astronomy (1931), solar batteries (1941), and transistors (1947). Then in 1952, Bell Labs would mark another groundbreaking technological advancement: The Audrey system a set of vacuum-tube circuitry housed in a six-foot-high relay rack that could understand numerical digits spoken into its speaker box. When adapted to a specific speaking voice, AUDREY could accurately interpret more than 97% of digits spoken to it. AUDREY is no doubt primitive by today’s standards, but it laid the groundwork for voice-dialing, a technology that was widely used among toll-line operators. (Remember those?)Ten years later, IBM unveiled its shoebox machine at the 1962 World Fair in Seattle. Like AUDREY, Shoebox could understand up to 16 words, including the digits 0 through 9. And when Shoebox heard a number combined with a command word (like “plus” or “total”), it would then instruct a linked adding machine to calculate and print the answer to simple arithmetic problems. Just like that, the world’s first calculator powered by voice recognition was born!

HARPY takes wing

Voice recognition began to take off as a field in the 1970s, thanks in large part to interest and funding from the U.S. Department of Defense and DARPA. Running from 1971 to 1976, DARPA’s Speech Understanding Research (SUR) program was one of the largest research initiatives ever undertaken in the field of voice recognition.

SUR ultimately helped created Carnegie Mellon’s Happy voice recognition system, which was capable of processing and understanding more than 1,000 words. HARPY was particularly significant due to its use of “beam search” technology, which was a far more efficient method for machines to retrieve the meaning of words from a database and better determine the structure of a spoken sentence. Indeed, advances in voice recognition have always been closely tied to similar strides in search engine tech — look no further than Google’s current dominance in both fields for proof-positive of this fact.

From recognition to prediction

By the 1980s voice recognition tech had begun to advance at an exponential rate, going from simple machines that could understand only dozens or hundreds of spoken words, to complex networked machines that could comprehend tens of thousands. These advances were largely powered by the development of the hidden Markov model (HMM), a statistical method that allowed computers to better predict whether a sound corresponds to a word, rather than trying to match the sound’s pattern against a rigid template. In this way, HMM enabled voice recognition machines to greatly expand their vocabulary while also comprehending more conversational speech patterns. Armed with this technology, voice recognition began to be adopted for commercial use and became increasingly common in several specialized industries. The 1980s is also when voice recognition began to make its way into home consumer electronics, like with World of Wonder’s 1987 “Julie” doll, which could understand basic phrases and reply back. 

Voice recognition goes mainstream

In 1990, we saw the release of the very first consumer-grade voice recognition product: Dragon Dictate, priced at $9,000 (that’s $17,000 in 2017 dollars). Following this, Dragon Dictate’s 1997 successor, Dragon NaturallySpeaking, was the first commercial voice recognition program that could understand the natural speech of up to 100 words per minute.

1997 also saw the release of BellSouth’s VAL, the very first “voice portal.” VAL was an interactive system that could respond to questions over the phone, laying the groundwork for the same technology powering the voice-activated menus you hear today when calling your bank or ISP. But after more than 40 years of advancement after advancement in voice recognition technology, developments in the field stalled out from the mid-1990s through to the late 2000s. At the time, voice recognition programs had hit a ceiling of about 80% accuracy in recognizing spoken words due to the HMM underpinning speech technology.

Google, Siri, and the voice recognition revolution

Apple’s iPhone had already made waves when it came out in 2007, as tech began to re-orient itself towards an increasingly smartphone-centric and mobile-first future. But with the release of the Google Voice Search App for the iPhone in 2008, voice recognition technology began to once again make major strides. In many ways, smartphones proved to be the ideal proving grounds for the new wave of voice recognition technology. Voice was simply an easier and more efficient input method on devices with such small screens and keyboards, which incentivized the development of hands-free technology.

But even more significantly, the design principles google laid down with Voice Search in 2008 continue to define voice recognition technology to this day: The processing power necessary for voice recognition could be offloaded to Google’s cloud data centers, enabling the kind of high-volume data analysis capable of storing human speech patterns and accurately matching words against them. Google’s approach was then perfected by Apple in 2011 with the release of Siri, an AI-driven personal assistant technology that likewise relies on cloud computing to predict what you’re saying. In many ways, Siri is a prime example of Apple doing what it does best: Taking existing technology and applying a mirror-sheen of polish to it. Siri’s easy-to-use interface combined with her sparkling ‘personality’ and Apple’s expert marketing of the iPhone helped make the program nearly ubiquitous.

The Potential Variables in Speech Recognition Software

“Correctness and accuracy are two different things,” says CallRail Product Manager, Adam Hofman. the difference lies in that correctness means completely “free from error” while accurate means “correct in all details” and “capable of or successful in reaching the intended target.”With speech recognition, this means that while the transcription may not be 100% correct (some words, names, or details might be mistranscribed), the user understands the overall idea of the chunk of speech that’s been transcribed. That is to say, it’s not just a jumble of random words–but that a cohesive concept can be interpreted from the text, in general. However, no two people are alike, and therefore, speech patterns and other deviations must be taken into account. Anomalies like accents (even those across English as native language speakers) can cause speech recognition software to miss certain aspects of conversations. The ways in which speakers enunciate versus mumble, the speeds at which they speak, and even fluctuations in speaker voice volume can throw speech recognition technology for a loop.

Regardless, most modern speech recognition technologies work along with machine learning platforms. Hence, as a user continues to use the technology, the software learns that particular person’s speech patterns and variances and adjusts accordingly. In essence, it learns from the user. CallRail’s voice recognition technology is used in conversation intelligence features like CallScore, Automation Rules, and Transcriptions.

The Benefits of Using Speech Recognition Software

Though speech recognition technology falls short of complete human intelligence, there are many benefits of using the technology–especially in business applications. In short, speech recognition software helps companies save time and money by automating business processes and providing instant insights into what’s happening in their phone calls. Because software performs the tasks of speech recognition and transcription faster and more accurately than a human can, it means it’s more cost-effective than having a human do the same job. It can also be a tedious job for a person to do at the rate at which many businesses need the service performed. Speech recognition and transcription software costs less per minute, than a human performing at the same rate, and never gets bored with the job.