Click to learn more about author Rachel Roumeliotis.

Artificial intelligence (AI) is now powering conversational commerce in retail, increasingly using chatbots to streamline and improve customer service. This can help with everything from answering customer queries and resolving issues to helping sell more merchandise through product recommendations. Voice-to-text translation is a crucial part of this understanding between humans and machines, and it’s getting more sophisticated by the day. 

While retail is one major application of this technology, we’ve really only scratched the surface of what’s to come. Here’s why: Audio is unstructured data. Transcription of this information enables us to more readily analyze it to provide useful business insights. Without translating this audio to text, companies that don’t have advanced natural language processing (NLP) capabilities in place to connect the dots between free-text and structured data – and most do not – are at a loss. 

To summarize, it’s hard to get the full picture of the consumer if we’re leaving out important parts of the conversation. You can imagine how detrimental this can be if we’re talking about applications in an industry like health care or pharmaceuticals. If you can’t link important medical history of a patient, whether stored in electronic medical records or physician’s notes, with concerns raised in a conversation during a telehealth appointment, treatment and outcomes could be compromised. 

Despite how successful voice-to-text translation has been across verticals, it’s not top of mind for potential users. In fact, recent global research from our company found that 18% of survey respondents indicated that immediate translations were the least exciting area for AI to develop in, preceded by 3D printing (11.7%), recommendations (13.3%), investment in technologies (15.9%), and facial recognition (16.8%). While voice-to-text translation may not have the allure of self-driving cars or virtual assistants, it does have real, practical applications for workers. 

In fact, nearly 30% of respondents from the same survey cited voice-to-text translation when asked what they would like AI to help them with at work. While the number is lower compared to other areas – knowledge delivery and assistance, software deployment monitoring and optimization, personalization, etc. – it can certainly improve employee workflow in day-to-day operations. This enables workers to focus on more high-value, business-critical tasks, which can pay off in dividends in terms of productivity. 

For example, widely used office suites like Microsoft offer voice dictation. Word’s speech-to-text feature enables users to speak words instead of having to type them on the keyboard. This helps capture stream-of-consciousness thoughts or simply reduces time spent manually punching letters. Voice transcription from customer service calls, conference audio and video, and other areas can be quickly tabulated and stored, affording listeners to be present in the moment and go back and review content to solidify their learnings at a later time. 

Another area voice-to-text translation can be particularly useful is to connect multilingual experiences. As the workforce becomes more distributed and global, a trend that’s been accelerated by the remote and hybrid work shift brought on by the pandemic, the need for language translation is becoming more prevalent. The nuances between speaking and writing can be difficult to decipher, but multilingual speech-to-text transcriptions can help level the playing field and foster a culture of collaboration among teams across the map. 

By capturing information in multiple formats – in this case, audio and text – we can gain better business insights. Fortunately, AI-enabled text-to-voice translation empowers us to do this in an automatic way that doesn’t require manual data entry, which is both time-consuming and prone to human error. While it’s still early days for voice-to-text translation, the potential to advance and grow is obvious. As NLP and other areas of speech recognition become more sophisticated, so too will the possibilities of what we can do in this area.