As we approach 2021, Michael Tansini at Speechmatics explains how automatic speech recognition can result in better conversations while offering a 360-degree view of customer interactions.
Remote working has seen an operational revolution take place in workplaces across the globe. With video conferencing now the norm, many companies are missing a trick by not extracting valuable information and data from these interactions. In 2021, having better visibility of the insights voice data holds will be essential. It will also offer countless benefits, including ensuring compliance and boosting operational efficiency. ASR will do the heavy lifting of capturing conversations so that valuable insights are not being neglected across the business.
2020 has shown that remote working has not been an easy transition for all, such as those working in a non-native language, and especially those with speech and hearing impairments. ASR can play a major role in helping to rectify these difficulties in the year ahead. The value of low-latency transcription in real-time becomes ever more urgent, to ensure people are involved and represented in conversations that concern them.
Consumer facing businesses will find that increasingly the first interaction with a customer will be virtual. This will mean that phone lines and chat rooms will become in effect virtual front doors and effective use of ASR in these scenarios can make the difference between someone stepping in or walking away. In 2021, businesses need to realise this and make sure that the reception they offer to new and old customers is inclusive, accessible and flexible – high quality speech recognition is one way to do this.
The enterprise ASR market has become heavily saturated especially with tech giants such as Amazon, Microsoft and Google jumping into the fray and offering the tech as part of their wider portfolio. Through a larger market share and well-established brand, these tech giants ultimately overshadow the ASR providers, despite not necessarily having better ASR solutions to offer.
The coming year will see a shift in this as the industry moves away from measuring only word error rate (WER), which up until now, has been the deciding factor in accuracy. Instead, we will start to see a shift where ASR engines will be measured on WER but also according to the engine’s ability to understand context and nuances in voice data, in various real-world use cases – think accents, dialects and colloquialisms easily being understood, and analysed.
The businesses which can differentiate themselves on this crucial point will dominate the enterprise speech recognition market going forward. As ASR moves from just a speech-to-text tool and starts to include capabilities such as automatic language detection, quality speaker diarisation, better punctuation, and even the ability to detect non-speech, the industry will see those companies that are able to better apply their technology to very specific use cases, improve customer interactions, while offering a 360-degree view of them, will come out on top.
