Speech Technologies in Unified Communications and Collaboration

Unified Communications and Collaboration describes the integration of voicemail, e-mail, SMS, VoIP and other forms of communication to allow each person within a company to communicate with the method most familiar or convenient to them. Advances in speech recognition and voice analytics technology mean that they can now be integrated into the UC&C stack. Before a company chooses a specific solution to satisfy its use cases, it is important that it understand each of the speech technologies now available, how they work and how they might integrate for maximum benefit.

Speech-to-Text
Speech recognition can be used to convert a voicemail message or other recorded audio into a readable text format. This has broad use as an accessibility tool for phone users with difficulty hearing, but can also be added to the suite of integrated forms of communication. Users can set up speech recognition to automatically record voicemail messages and forward the result to their e-mail or another preferred means.

Speech recognition is already built into many devices and operating systems on a basic level. Some of these implementations are adaptable enough to recognize the user’s specific speech patterns or accent for improved results. These default implementations are not ideal for unified communications as they are typically intended for live dictation rather than transcribing recordings other than voicemail. A dedicated solution can be fine-tuned to your organization’s needs and configured to transcribe recordings of conversations in whichever format is most convenient for your use. It can also directly store or forward the results to other text-based destinations according to your settings.

Text-to-Speech
Conversely, text-to-speech allows users to listen to messages while on the move or working on another task. This function is natively supported on most modern mobile phones, either under a dedicated speech setting or through accessibility features. Browser plugins can be used to automatically read text from e-mail messages aloud. These approaches yield acceptable results but have limitations in word recognition and speech delivery. Again, a dedicated unified communication solution is better-equipped to handle this use case.

Voice Analytics
Voice analytics or speech analytics generate usable data from live or recorded audio. Software can be used to analyze phonetics or specific words, depending on whether faster processing or faster queries are prioritized. When a recognized pattern occurs, it can be used to create alerts in real time or as a post-processing step to sort or categorize recorded audio files.

This technology is frequently used to analyze incoming calls from customers to find calls in which they sound dissatisfied and to identify improvements that can be made in the customer service process. It is also used for logging calls and retrieving information about them during review. The same analysis can be applied to internal conversations and recorded voicemail messages to identify areas where confusion frequently occurs, areas in the communication workflow which can be optimized or simply for archiving important conversations to be reviewed or referenced in the future.

As speech-to-text and text-to-speech transcription, speech recognition, voice analytics, and AI continue to advance, they will open up myriad new opportunities in the field of unified communications. In turn, this will give organizations the chance to derive value from these human-to-machine and machine-to-human interactions. But as with any incipient technology, it is in their best interest to fully understand the technology and its possible ramifications prior to pursuing implementation.

2018-04-23T18:11:32+00:00