Four ways to transform Customer Contact Center Interactions

Globally, contact centers are viewed as cost centers – while reducing operations cost of the contact center is the biggest challenge, business leaders are also waking up to the fact that end customers can be delighted over every single interaction leading to customer loyalty and new revenue opportunities!

Contact center automation Uniphore

Let’s take a look at some of the key contact center industry facts and figures:

  • Companies spend up to $15,000 to add a new call center agent to their staff with an average retention of 6 months and a 40 percent annual churn.
  • 15% to 45% of calls handled are of the same nature and can possibly be automated but for the different languages and dialects that customers use when they call.
  • The global cost of fraud and identify theft is now estimated to be over $200 billion per year
  • Customer expectations are reshaped by faster response over social media, and they demand better service through contact centers:
      1. Lower “Average Speed of Answer (ASA)”
      2. Better “Average Call Handling Time (AHT)”
      3. Better “First Call Resolution (FCR)”

BPO Uniphore contact center automationHow do you balance above challenges and still manage to delight the millions of end customers who call you regularly?  Can technology help in seamlessly solving customer issues and at the same time setup a platform for improving customer loyalty?

Here is where Uniphore’s Contact Center Automation solution can help you. Uniphore’s solutions allow any machine to understand and respond to natural human speech, thus enabling humans to use the most natural of communication modes, speech, to engage and instruct machines. Our solutions have been silently revolutionizing the customer contact center interactions in the following 4 ways:

  1. Voice Biometrics Authentication UniphoreCustomer Authentication – Voice Biometrics provides next generation security for enterprises to authenticate its customers. To gain access, the customer has to authenticate using their voiceprint which is as unique as their fingerprint and is impossible to forge. It can be deployed over IVR and data-enabled applications to provide support across all mobile devices and platforms. Example: Instead of the agent asking 4 to 6 authentication questions to a caller over 3 to 5 minutes, Uniphore’s voice biometrics solution can authenticate the caller in less than 15 seconds thereby saving time, effort, cost and also guaranteeing complete security.
  2. contact center call resolution uniphoreIssue Resolution – Speech Recognition is the ability of an application to understand and carry out spoken commands. With speech recognition integration, enterprises can enable natural, human-like conversations and satisfying interactions with customers. The IVR system will understand customer’s speech commands and execute appropriate transactions instantly. Example: A typical caller is struck with dial-pad based IVR router for at least 60 to 180 seconds to get through the appropriate help line. Uniphore’s Speech Recognition solutions enable callers to directly ask their query and have human like conversations.
  3. Fraud detection contact center automation uniphoreFraud Detection – Audio mining embraces speech recognition technology to extract insights from prerecorded voice streams. This insight can then be used to classify calls, trigger alerts/workflows, and drive operational and employee performance across the enterprise.  Example: Uniphore’s Audio Mining solution can help process thousands of hours of audio data and identify the word ‘fraud’ and help to know why the customer or the service executive used the word and take necessary actions immediately.
  4. process compliance contact center automation uniphoreProcess Improvement - Aforementioned audio mining solution can also help in staff training and process improvement.  Examples:
      1. Usage of “I don’t know” as a key phrase to identify crucial knowledge gaps and develop targeted training.
      2. If the words “bill” and “overcharge” start occurring together often, company can quickly retrospect on their billing procedure and course correct as required

Uniphore Software Systems is a pioneer in the speech based solutions market and has built voice based solutions in 14 languages and 100+ dialects.   Would you like to know more? Write to me at and I will be more than happy to send you the relevant details!

Posted in Audio Mining, BPO, Contact center, Natural Language Processing, Speech Recognition, Voice Biometrics, Voice Solutions | Tagged , , , , , , | Leave a comment

Field force automation – Tips & Tricks

Much of the success of an enterprise, a service provider in particular, depends on customer satisfaction. Field force automation is one of the ways of ensuring it.  The goal is higher quality, more meaningful engagement, better information, better timing, and above all, a more proactive and responsive company culture. A recent study reports that a 5 percent improvement in customer retention could increase profits by as much as 85 percent. CIOs everywhere are trying to find innovative ways to enhance customer loyalty, increase revenue, and improve productivity.  Consider the following scenarios:

NBFC Field Force Uniphore

       Agribusiness Field Force Uniphore

“Operations team with 800 field agents incur a 40% productivity loss due to back & forth communications” - Leading BFSI Company

“Logistics division has built 1750+ strong field force across all major Tier I and Tier II cities – Merchandize returns effects up to 40% loss in margins” - Large eCommerce player

“Clearing & Forwarding Agent firms deploys field agents to engage with lakhs of farmers across 12 remote mandals – Time delays between CFAs, dealers, retailers and farmers lead to more than 15% revenue loss” – Major Chemicals & Fertilizers AgriBusiness

Needless to say, field agents have to be managed tightly to ensure higher productivity, reduced operational expenses and also improve customer experience. And this is precisely what, mForce, Uniphore’s mobile solution does. mForce is a feature rich software that integrates multilingual speech recognition and voice biometric capabilities with standard mobile data (GPRS) technology to help:

    • Optimize the field agents delivery routes
    • Automate delivery status and acknowledgement
    • Simplify payment processing and receipts
    • Receive customer feedback information
    • Track field force in real-time

To know more, join our upcoming Webinar, “Field Force Automation – Tips & Tricks” (27th March; 3PM IST) and learn how to boost your field force productivity by 40%!




Posted in Agriculture, eCommerce, Enterprise Mobility, Field Force Automation, Logistics | Leave a comment

How to build Enterprise Mobile App in 24 hours?

Enterprise Mobility Scenario:

enterprise mobility uniphoreMobile business apps market is set to reach $53bn in 2017, says a leading market research firm. It is also predicted that 200 million mobile workers will use at least one business-focused app this year. Over the next few years, both the number of mobile workers using enterprise mobile apps and the number of apps available will grow rapidly.

Factors to consider before you build Enterprise Mobile App: 

While consumer apps are built on cutting edge technologies and provide best-in-class GUI, negligible latency, multiple hardware integration features, enterprises apps tend to be slightly sluggish!  In their defense, enterprises are weighed down by these factors:

  • Integration with multiple corporate back end systems is never easy – as they say, no chain is stronger than the weakest link.  One cloggy system can potentially bring down the whole ecosystem.
  • Offline functionality – be it a senior executive flying in a plane or a travelling field sales person in remote location, we have enough scenarios where the data connectivity does not exist.
  • Business functions – each functional areas will have their own requirement and the complexity keeps increasing with addition of newer scenarios
  • Security – as much as BYOD is consumer friendly, it is also a nightmare for IT personnel. Impact of lost devices, employee attrition, identity theft is very high. While most of the data & applications in laptops can be controlled through software like Microsoft’s Active Directory Federation Services (AD FS), enterprise mobility is a different world.   The onset of personal devices usage in workplace make it complicated to weed out malicious code and “containerize” content
  • Device - plethora of devices, vendors, form factors, Operating Systems (not to mention versions), configurations (RAM/CPU) doesn’t help to “build once, deploy anywhere, run anytime” unless you have enterprise app store capabilities
  • Time to market – in a fast paced world, smart mobile users are so used to “Download, Install & Run” in a matter of few seconds without any glitches.  Employees and customers expect the same pace for all enterprise software as well – this gives a very small window for the development lifecycle.

build enterprise mobile app fast uniphoreHow to build Enterprise Mobile App in 24 hours?

Now, how can CXOs and IT Heads possibly respond to satisfy all the Business Heads who demand rapid deployment of mobility solutions to match and exceed their competition? Here is where Uniphore’s Mobile Control & Access System (mCAS) comes to the rescue.  mCAS, is Uniphore’s Smart Mobility application development and management platform. Enterprises leverage this platform to quickly build and deliver customized, cross-device applications to customers.

Features of mobile Control & Access System (mCAS):

    • Application Development Framework: Drag-and-drop GUI interface for developers to create applications across – no coding required
    • Mobile App Management (MAM): Remote Application Scheduling and Deployment , User Groups Configuration, App Users Management, Policies Configuration, etc.
    • Mobile Device Management (MDM): Device Location Tracker, Security Policy, Remote, Wipe, Statistical Reports
    • Mobile Expense Management (MEM): Proactively monitor and track mobile data and application usage
    • Speech Service Integration: Speech Plug-ins. Including Speech Recognition and Voice Biometrics
    • Customized Report Generator: Dynamically configure any and all reports required

Please do write to me at for a demo of mCAS platform.

Posted in Enterprise Mobility | Tagged , , , , , , , , , , , | Leave a comment

Hear Watson, hear.

In 2011 IBM’s Watson Computer won the American quiz show Jeopardy. It did so on live TV, beating two of the all time great (human) contestants of the show. In the Indian context, this would be a bit like a computer going on Kaun Banega Crorepati, interacting with Amithabh Bachan, and becoming a Crorepati.  Are we going to see that happen some day?

Though Watson did play the game like a human contestant, thinking about the next move, pressing the buzzer etc., a little known fact is that Watson was fed the questions in text form rather than speech. Some people think this made the whole show a gimmick. After all Google can consume text too. So what was the difference between them? In the simplest terms, Google returns the most popular page containing a phrase. Watson “understands” the phrase.

So if you ask Google a question like “How many cows are there in India?” Google will return all the pages on the entire internet, containing the word Cow and India. A few of those pages will have the real answer. The top most link returned may or may not have the most current information, as the popularity of the page makes a difference to Google. Watson on the other hand will return a number. The most current known number.

Try it out here – – which uses similar underlying tech.

So what did IBM show us? They showed us, a system can be built that can answer questions. Real answers not just pages of links. Answers to anything. They showed that it required 2880 processor cores with access to 15 terabytes of human knowledge. The whole setup was the size of a large bedroom. But that was two years ago, and we all know how things change these days.

Today Watson is the size of a pizza box. And IBM is expecting it, to pass the US Medical Licensing Exam any day now. It is just a matter of time, before we have a doctor sitting inside our mobile devices. And many other domain experts there after. So will Doctors become irrelevant? Definitely not. They will just become more effective at what they do. After all game shows haven’t disappeared.

Speaking of which, are we going to see a computer win Kaun Benega Crorepati? Will it happen in Hindi and Tamil? Most definitely yes.

NLP or Natural Language Processing is the technology that allows Watson to “understand” the text. It is the technology that both the Uniphore Product and R&D teams will continue enhancing this year. Uniphore is a pioneer on local Indian languages in both Speech Recognition and NLP, and has helped businesses reach out to 3,500,000 end customers in 14+ languages and 100+ dialects.

Write to Sharun Shanthosh (AVP – Product Engineering, Uniphore Software Systems) at to understand more about our Speech Recognition and Natural Language Processing technology.

Posted in Natural Language Processing, Speech Recognition, Voice Solutions | Leave a comment

mHealth: Voice-based solutions for chronic disease management

Chronic diseases are long-term diseases that develop slowly over time, often progressing in severity, and can often be controlled, but rarely cured. They include conditions such as cardiovascular diseases (heart disease and stroke), cancer, diabetes, arthritis, back problems, asthma, and chronic depression. Worldwide, chronic diseases have overtaken infectious diseases as the leading cause of death and disability. India is no exception to this trend. In fact, about 50.9 million people suffer from diabetes alone in India, making it the ‘Diabetes Capital’ of the world.

voice based disease management

In treating chronic diseases, medical practitioners recommend that clients be meaningfully involved with the disease and its symptoms, soliciting them to be active partners in managing their condition, rather than passive recipients of care. Moreover, best practices suggest that support for individual self-management should be an integral feature of the health care organization.

Clients need a range of support services to become effective self-managers of their chronic conditions. Five kinds of self-management and training support services are key:

    • Information and education about the disease, its causes and effects, strategies to forestall its progress and prevent disease complications, and effective treatments
    • Training in skills to manage the disease day-to-day. These range from technical skills, such as monitoring blood sugar levels, to problem-solving and coping techniques. Improving client self-efficacy – the confidence to actually do what needs to be done – is also an important skills objective.
    • Behaviour modification programs to help clients change the behaviours that exacerbate their condition or increase the risk of complications. Examples include smoking cessation and physical exercise programs.
    • Counseling, advice and supportive services to help clients learn how to cope emotionally with their conditions
    • Linking with all available health and social resources in the community.

Furthermore, the final key to self-management support are the follow-up measures that help providers and clients stay on track with the care plan, and day-to-day self- management activities. These follow-up measures include appointment and check-up reminders, checks on client compliance with their care regimen (including medications), and ongoing help with self-monitoring.

In the developed world, most of these health promotion activities are delivered in community settings, such as hospitals or workplaces. In India, however, the infrastructure to support such high-touch services is not as easily available. Therefore, it is critical that healthcare organizations leverage the deep penetration of mobile phones in order to supplement the human support services they provide.

For example, Uniphore works with a rural healthcare organization in Tamil Nadu focused on treating cardiovascular diseases. The business was struggling to meet each patient often, as the costs of contact were so high. Therefore, in order to consistently and meaningfully support their patients, the company elected to create an automated, voice-based support system.  Once a patient is entered into the system, they are put on a ‘schedule’ of calls, in which the system automatically reaches out to them in order to support them in their disease management. For example, early on in the disease, patients get calls on a daily basis with suggested behavior modification in terms of lifestyle and diet. They are also educated about various complications with the disease they are experiencing, and the choices they have for dealing with those complications. Most importantly, the system also engages in interactive conversations with the patient to ascertain their levels of medication and treatment compliance. It simple enquires whether or not patients have taken their medication, and when they have not, it helps to solve the problem. All of these calls are conducted in various local languages as required. Additionally, using Speech Recognition, the system will capture the responses of the patients, converted them to text, and store them in a backend MIS.

mHealth AllianceFor our innovative work in disease management and awareness building in the healthcare domain, Uniphore has recently been accepted as a member of the global mHealth Alliance. We look forward to continue to tackle the challenges of healthcare delivery in India and across the world, leveraging speech technology as a tool for meaningful communication between organizations and their patients.

Please write to me at for more details about our voice based mHealth (disease management) solutions.

Posted in Healthcare, mHealth, Voice Solutions | Tagged , , , , , | Leave a comment

Woes of Last Mile Delivery

The world’s largest online retailer,, was quick to realize that the value was shifting from “what” was delivered to “how” it was delivered – Kiva Robots, AmazonFresh and futuristic drone delivery are all part of solving the woes of last mile delivery.  While the $62 billion retail giant keeps innovating, our Indian online retailers like,,, HomeShop 18, and have their own set of unique battles to fight.

Woes of Last Mile Delivery

The Indian demography and preferences complicate the “last mile delivery” in the following ways:

  • Cash on delivery – 50% of online orders are on COD which means delivery agents carry about 15,000 to 25,000 rupees in cash at any time.  This makes the whole process risky, expensive and laborious!
  • Merchandise return – buyer remorse is a global phenomenon but it is more acute in India leading to reverse logistics, multiple deliveries and effecting up to 40% loss in margins
  • Postal address – a minimum of 10% fuel is wasted on “false miles”; most recipients have to be called multiple times for nearby landmarks & further navigation guidance and this has a direct impact on delivery agent productivity.  This leads to dependency on local field agents as well as inability to utilize field agent of one area in other areas

Assuming that the quality of the product(s) delivered is met, customer satisfaction is directly dependent on two main processes in last mile delivery:

  1. Order tracking – back end call center agents will monitor consignment from supplier to warehouse to local hub, assign orders, guide agents to locations, get customer feedback and report
  2. Order delivery – field agents will collect package(s), locate address, deliver package, collect payment and acknowledgement as required

The whole process has to be managed tightly to ensure higher productivity, reduced operational expenses and also improve customer experience. And this is precisely what, LastMile 360, Uniphore’s mobile solution does.  LastMile360 for e-commerce and logistics industry, is a feature rich software that integrates multilingual speech recognition and voice biometric capabilities with standard mobile data (GPRS) technology to help:

    • Optimize the field agents delivery routes
    • Automate delivery status and acknowledgement
    • Simplify payment processing and receipts
    • Receive customer feedback information
    • Track field force in real time

For more information, please write to me at and I can explain in detail about how we solve the woes of last mile delivery!

Posted in eCommerce, Field Force Automation, Logistics, Retail | Tagged , , , , | Leave a comment

How technology can solve the major challenges in Microfinance

Last week, experts, policy makers, practitioners, researchers and journalists gathered together for the two-day Microfinance India Summit in Delhi. There were a plethora of sessions covering a wide variety of topics, specifically the key constraints limiting the sector’s growth and expansion.

Two key areas emerged as challenges:

1. Effective on April 1st 2014, margin caps may not exceed 10% for large MFIs and 12% for others. In the medium term, this is expected to put significant pressure on MFIs to reduce operational costs.

2. India is one of the world’s least financially aware nations, as reported by a global survey conducted by VISA. With less than 35% of Indians’ qualifying as financially aware, potential MFI customers are less likely to take out a loan, and to pay it back responsibly.

Although these two problems are significantly different, Uniphore’s experience shows that technology can play a meaningful role in alleviating both of them.

If MFIs need to reduce their operational costs, an important step in this direction is cutting the number of branches from which they physically operate, and enabling their mobile field agents to operate over larger geographies. However, in order for this to happen, it is critical that the MFI be able to remotely support and monitor these field agents throughout their daily activities. A mobile fieldforce application, like Uniphore’s mForce, can help MFIs do just that.

The solution automates both data collection as well as loan disbursement and payment collection. All of the information that is traditionally carried out on pen and paper (such as customer onboarding forms, social impact forms, etc) can be captured on a smartphone. This not only reduces the costs of pen-and-paper processes (printing, courier, and data entry), but also eliminates much of the delays in the loan process, thus positively impacting turnaround time and customer satisfaction. The seamless exchange of data between the field and the head office will also bring down operational costs through faster customer identification and error resolution – slashing the number of trips an agent has to make to accomplish these tasks.

The issue of low financial awareness can also be solved through a solution that enables MFIs to more regularly train their clients. People cannot be expected to retain extensive information in one sitting alone, and thus they require a more personalized communication tool that regularly trains and reminds them about how to use credit, budgeting skills, etc. While an MFI cannot consistently do these trainings in person, they can leverage widespread mobile penetration to create a solution. For example, Uniphore’s VoiceNet platform has been used to conduct interact quizzes for an Indian MFI. Using Indian language speech recognition, the system can call any basic feature phone, and ask quiz question related to financial literacy. Answers are recorded and tracked by the organization, so they can use these insights for analysis and further trainings.

Experts and practitioners at the Microfinance Summit equally acknowledged the critical need for technologies of these kinds to enable more efficient and meaningful interactions with both field agents and customers. The future will surely see high levels of technology adoption as MFIs strive to meet the double bottom line.

Posted in Uncategorized | Leave a comment

Why Indian Enterprises Prioritize Software Spending

Despite challenging economic conditions, the enterprise software market in India is projected to reach $3.92 billion in 2013, a 13.9% growth over 2012 revenue of $3.45 billion. Indeed, in 2013, India will be the fourth largest enterprise software market in Asia/Pacific. Gartner, a global technology analyst firm, attributes this growth to the following trends:

“Growing maturity of Indian users is an important driver for overall growth. Compounding the demand is the ongoing tendency for greater customer services along with the continued drive for IT cost savings, as well as the incorporation of emerging technologies into solutions, such as mobility, social, cloud and business process management. These new technologies, and demand for agility, brings in additional urgency, and demand, for IT investments.”

As Gartner points out, Indian enterprises are investing in transformative technology solutions primarily in order to achieve 2 things:

  1. Lower Operational Costs
  2. Better Customer Service.

Mobile applications for fieldforce or salesforce automation are one great example of the kinds of solutions that are enabling these two benefits. For example, Uniphore worked with one of India’s largest NBFCs, which employs over 6,000 field agents across India in sales and collections. These agents typically travel extensively in their territories to acquire new customers and service existing customers. Traditionally, they filled out the appropriate paper form for each transaction, and mailed the completed paperwork to their administrative headquarters at the end of each day. The time it took to fill out the forms, deliver them to HQ, wait for HQ to complete the processing and finally receive the administrative revert, was hampering field agent productivity and negatively affecting customer satisfaction.

The solution Uniphore provided essentially eased each step of the sales process. For each operation – lead capture, followup, document collection, feasibility reporting, etc. – the agent opens the relevant application on their phone, and enters the appropriate information. Each form can be filled out through either text or voice inputs, depending on the preference and literacy level of the agent.

Moreover, field agents also have access to their record of activities and reports on the status of each lead. When dealing with customers, the field agents can immediately handle queries, as their interaction history with that customer is easily accessible. Furthermore, every time a customer does a payment or has a meeting with the agent, they get an SMS as a confirmation of the engagement.

Using this same platform, the company builds and disseminates new forms, alerts, and updates, and sends them to particular agents or categories of agents. Furthermore, if a customer query comes into the company’s hotline, the administration can route the query to the agent who is responsible for that particular customer, enabling the agent to respond immediately.

The entire Mobility Solution is hosted on the cloud, allowing the business to embrace a pay-as-you-grow model with limitless scalability. Using this solution, field agents now achieve a lot more in a single meeting with a customer because the company can now process information and revert with decisions to its agents in real time. This has reduced the average sales cycle drastically, allowing agents to close a sale in just two meetings, down from four meetings from before. This reduction in customer visits directly translates to a decrease in travels costs for the business by as much as 50%.

A more efficient service provider always translates into more satisfied customers. In this case, the ability for the administrator to route queries from their overall system directly to the agent on their mobile phone, means that the agent can respond to the customer more immediately. This has enhanced both customer service as well as customer loyalty to the company.

With benefits like this, no wonder the Indian market for software solutions continues to rise dramatically.


Posted in Uncategorized | Leave a comment

How Enterprise Mobility can Transform Agribusiness

As enterprise mobility becomes increasingly mainstream, we often hear case studies about its impact in banking, insurance, retail, and manufacturing. However, in countries like India, where 70% of the country are involved in the agriculture industry, it is critical that these industries also be disrupted with innovation. These solutions will vary in functionality and platform from mainstream mobility solutions, as this industry and its stakeholder primarily use basic feature phones, and speak a variety of local languages. Nevertheless, huge benefit can be gleaned from these solutions when done appropriately.

For Indian agribusinesses, there are huge limitations posed by traditional methods of outreach in field operations. Farmers are spread across such a large area that individual visits are highly expensive and time consuming. Unfortunately, since regular and intimate contact with the farmers is crucial, these visits seem like an absolute necessity. To complicate the dilemma, agents often plan their visits based on personal convenience as opposed to the farmers’ requirement.

This is why experts agree, it is imperative that agribusinesses take advantage of mobile technology in servicing their farmers. AgriConnect, Uniphore’s voice-based mobility solution, has already helped some of the leading Agribusinesses in the country reach out to their farmers efficiently and at a fraction of their previous costs.  AgriConnect uses Speech Recognition technology to send automated calls to farmers in their own regional language, engaging them in dynamic human-like conversation. The system is able to answer all the farmers queries, and collect important information from them, which is linked to their profile on the web-based platform at the head office This profile enables AgriConnect to send the farmers timely and automated updates. Needless to say, delivering and receiving information through AgriConnect is a whole lot faster than by traditional, manual methods.

Watch this video to learn how AgriConnect works.

Posted in Uncategorized | Leave a comment

From Audio to Action: Demystifying Speech Recognition

As speech recognition starts to live up to its promise, more and more businesses are exploring how they can apply this technology to their operations. One of the first questions we always get is, “How does the technology work?” Therefore, below we have described the process of how exactly audio streaming can be understood, converted to text, and applied to automate and enhance communication. Enjoy!

The first step of speech recognition is turning a continuous audio stream into the basic sounds of the language, which are called phonemes.  Every language has different sounds, or phonemes, from which all the native words are built. In order to identify the sequence of phonemes in continuous speech, the computer divides up the incoming audio into short time-slices called “frames”.

For each frame, the computer measures the “strength” of pre-defined frequency bands within the overall range of speech. Thus, each frame is converted into a set of numbers, one number per frequency-band. The recognition engine uses a reference table to find the best matched phoneme for a given frame. This table contains representative frequency-band-strengths for each of phoneme. This is why users are asked to record a profile when using the speech recognition platform for the first time – it needs to know how you pronounce each of the phonemes, so as to get the best possible entries in the look-up table.

Next, the system must translate phonemes into words. The recognizer uses a lexicon which contains all the words it knows about together with their pronunciations – each pronunciation is described using phonemes. For example, the pronunciation of the word “cat” has three phonemes: ‘k’, ‘ae’, ‘t’. Some words have multiple pronunciations and the lexicon has these too, for example “either” has two common pronunciations – “ay dh ax” (eye-th-er) and “iy dh ax” (ee-th-er).

As the recognizer moves along the sequence of phonemes, it looks for words “hidden” in the sequence. It is allowed to create overlapping sequences of words, as different words and phrases may share the same pronunciations. For example, the lyrics from this song, “life is butter melon cauliflower” sounds the same as “life is but a melancholy flower”.

After the engine has identified its “candidate” word sequences, it must sort out which is the correct one by using language modeling. A language model describes speech patterns in terms of words which are likely to be seen together. All language models start from collections of the things people say or write in a particular context. For example, if you want to create a language model for the Times of India, you might compile a year’s worth of editions and generate the relative counts for all the two and three word sequences you find in that collection. Similarly, if you want to create a language model for a medical specialty, you might gather transcripts of reports in that specialty and compile the relative frequencies of all the two and three word sequences in those transcripts. Models constructed in this manner represent a kind of “average”, since they reflect the combined usage of a lot of users within a given field.

The language model helps us sort between the competing sequences of words from the conversion of phonemes into possible words and phrases. For example, suppose the recognition, thus far, has yielded two possible fragments — “over there” and “over their”. If the next word identified is “heads”, then the language model would help the engine choose “over their heads” as opposed to “over there heads”.

Now that the audio has been converted to text, the world is your oyster. This text is yours to be searched, shared, analyzed, or visualized – depending on what your want to achieve.


Posted in Uncategorized | Leave a comment