United Telecoms launches real-time call transcription, sentiment alerts for SA contact centres and SMEs

A United Telecoms customer service agent handles a live call, with real-time transcription and sentiment alerts running in the background. (Image: United Telecoms)

---

United Telecoms, a South African provider of PBX, VOIP, unified communications and contact centre solutions, has added real-time call transcription and sentiment alerts across its United Voice Cloud ecosystem, spanning its call centre software and cloud PBX platform, making a capability once reserved for large-scale contact centres available to any South African business that handles calls on a regular basis.

The new capability turns live voice calls into searchable, structured text the moment a conversation happens, giving South African businesses a practical way to act on what customers are saying without risking losing information the second the call ends.

Where e-mail and WhatsApp conversations have long been mined for insight, voice calls have typically sat as unstructured audio in a recordings folder, reviewed only when there's a complaint to investigate or a compliance box to tick. United Telecoms' new real-time transcription and sentiment analysis tools are designed to close that gap.

Contact centres generate enormous volumes of customer conversations every day. Industry research on contact centre quality assurance shows why so little of it gets used. The average QA analyst manually reviews only a small fraction of total call volume, often as little as 1% to 3%, leaving the vast majority of voice calls completely unmonitored.

Smaller South African businesses face the same problem without even having a formal QA process to fall behind on. Without a dedicated analyst, most calls on a standard business phone line simply go unrecorded and unreviewed.

Whether it's a formal contact centre or a small team handling calls on a standard business line, South African businesses handle high volumes of sales calls, support queries and conversations in regulated industries such as insurance, lending and debt collection. For all of them, that gap can mean missed coaching opportunities, inconsistent service quality and customer complaints that go unnoticed until they've already escalated.

Instead of relying on manual note-taking or professional human transcribers to reconstruct what was said after the fact, businesses can now capture call transcripts as the conversation happens, then route that written transcript wherever it's needed most:

• Faster agent performance reviews: Supervisors can search call transcripts for key phrases instead of listening back through hours of call recordings, making quality assurance and coaching far less time-consuming.
• Real-time sentiment alerts: Built-in sentiment analysis flags calls that are heading in a difficult direction, giving supervisors the chance to step in on live calls before a frustrated customer becomes a lost one.
• Searchable, compliant records: Wherever transcription is switched on, conversations are stored as plain text alongside the audio, giving businesses a written record that's easy to search, audit and retain for compliance purposes.
• Better follow-ups: Sales and support teams can pull accurate call transcripts into their CRM or help desk instead of relying on memory or hurried notes when making follow-ups.
• Support for multiple languages: Because transcription runs through providers like OpenAI, Deepgram, Google and AWS, businesses can choose the provider best suited to the languages and accents their customers use, rather than being locked into a single-language system.

"Sentiment analysis is only useful if it understands how your agents and customers talk," said Ryan Lumley, Director at United Telecoms.

"A model trained on generic data might not know or understand South African accents, industry terms or what a frustrated customer actually sounds like on a call. We tune that with each business directly, so what a supervisor sees on their screen reflects what's really happening on the call." 

When a call starts and call recording is enabled, the platform automatically opens a live transcription session. Stereo audio is streamed in small chunks to a speech-to-text provider, currently supporting OpenAI, Deepgram, Google and Amazon Web Services, and converted into text in real-time.

The resulting transcript is published to a destination the business defines, whether that's a CRM, help desk tool, analytics dashboard or a custom-built application. Because transcription can be switched on at the queue or team level rather than system-wide, businesses can activate it exactly where it adds the most value, for example, a single sales extension or across a sales team, VIP client queues or support desks where every call might need to be pulled up later, without transcribing every call across the organisation.

The feature builds on United Telecoms' existing AI voice agent platform, which handles live conversations with customers. Together, the two tools give businesses AI support during the call and a written account after the call ends.

Existing United Voice Cloud customers, on both the phone system and omnichannel contact centre software suite, can activate transcription on selected queues or extension groups without new hardware.

From there, United Telecoms' local team of experts configures the speech-to-text provider, sentiment thresholds and integration destination for each deployment, acting as a single point of contact for set-up, integration and ongoing support.

How to get started with real-time call transcription

South African businesses looking to turn call recordings into usable data can activate real-time transcription through United Telecoms' omnichannel contact centre software or cloud PBX platform today. Contact United Telecoms to discuss pricing, integrations or how transcription and sentiment alerts could fit into your existing set-up.