Tips for Small Businesses That Want to Use Automatic Speech Recognition (Infographic)

Automatic speech recognition technology (ASR) and its complementary technology known as IVR (interactive voice response) are both systems you’ve almost certainly used before in a personal or business context.

Apple Siri

These are the technologies that allow machines to interact with humans through audio communication and anybody who’s ever used an automated telephone customer service system, a voice powered text editing program or a smartphone interface like Siri on the iPhone has come into contact with one kind of ASR/IVR software or another.

The possible business applications of these systems are numerous and your company could definitely benefit from using ASR technology in a number of contexts.

However, before we get down to some tips for that, let’s first do a quick overview of how automatic voice systems work and “learn” to interact with us.

The Essentials of ASR

The basic mechanism of nearly all ASR technology consists of the following steps:

  1. You talk to an ASR interface
  2. The software inside the interface creates a raw wave file of what you said
  3. The wave file is cleaned up with background noise reduction and volume modulation
  4. The filtered wave file is broken down into the basic building-block sounds of works, known as phonemes. Sound blocks such as “ka” “sh” “wh” and “t”. English has 44 of them.
  5. Each phoneme is analyzed in sequence as part of a long chain of sounds in your wave form and thus the ASR software intuits whole words and sentences from this analysis.
  6. The ASR system can then meaningfully respond to you.

Some ASR Types

There are two basic kinds of ASR systems we’ve all dealt with. These involve “directed dialogue conversations” and “natural language conversations”.

Directed dialogue systems are very, very common in the front end, customer service landscape and involve the automatic telephone customer service systems many companies use. With these limited systems, the ASR software only recognizes words you say from a limited choice of selections and responds verbally to only those words.

Natural language conversations on the other hand are far more complex and involve something that closely approximates real human conversation. The more sophisticated the systems, with a larger vocabulary, the more “human-like” their ability to understand and speak to their human users. The iPhone’s Siri is a fine example of such a system.

How Natural Language ASR Works

Natural Language ASR systems also learn to respond in conversation in a way that’s “similar” to human word comprehension. Instead of scanning their full vocabulary (typically about 60,000 or so words) they respond to certain “tagged keywords” that give them context enough to understand what you’re saying to them.

Thus, if you tell a natural language ASR that you want the “weather forecast”, the tagged keyword “forecast” will allow it to understand that your other word is “weather” and not “whether” which sounds identical but has a completely different meaning.

How ASR technology is “Taught” to understand better

ASR systems can either be taught by human guided “tuning” or via autonomous automatic learning.

Tuning is the much simpler process of ASR “training” and involves human programmers going through the software’s conversation logs for new words and phrases that they can then include in its dictionary for future use.

Automatic learning on the other hand involves programming the software so that it learns to capture, recognize and adopt new words as it hears them and then use them in the future for better conversation comprehension.

What ASR Means for your Small Business

Front End (customer interaction)

Now that you’ve gotten a clear overview of how ASR works, you’re probably thinking of some uses for it in your business.

For starters, there is the most basic cost saving client side use of ASR that your company might try out if it has a lot of simple customer queries to deal with. This involves installing an ASR telephone system to field phone calls with basic questions from businesses or for use as a customer feedback request tool.

Companies who offer ASR services, such as West Interactive, which created the detailed infographic that accompanies this post, have different solutions for different business budgets and needs.

ASR infographic by West Interactive

Back End (workflow management)

ASR software can also be used in numerous ways on the back end of a small business for the sake of simplifying workloads and the flow of tasks. With voice controlled text editing software, a person can more flexibly multitask and coordinate projects that involve going through a lot of written material.

Furthermore, dictation of invoices, memos and other company documents is a lot faster than typing them out once one gets used to workflow related ASR technology.

In essence, look at your company and where it could possibly benefit from ASR systems in terms of cost reduction and workflow simplification.

Cover photo credit: AKuo Yeh / Flickr


Leave a Reply

Your email address will not be published. Required fields are marked *