Speech to Text and vice-versa

Smart Language Handling and Analysis

Voice to Text has been around for decades and its always been 'ok', never good, not really bad but just ok.


Large Language Models (LLM's) bring a whole new dimension to voice transcription, being able to not only transcribe hard to follow accents, but some of the larger models can transcribe in 30 different languages.


Services are available online to achieve this at a reasonable cost, but there's a huge risk here especially with sensitive transcripts like call recordings. We can easily host these technology on your site in a closed system for all the privacy and security benefits.

Hero Image

Voice to Text

We can easily automate the transcription process to digitise call recordings and other audio files and make them available in various formats for you to access. We can further expand this by performing text analysis on the transcripts to verify compliance with required scripting and regulations. We can also provide exception reporting for transcripts that mention or do not mention specific keywords or phrases.

Text to Voice

Text to voice, yes, we can do it both ways. We are often asked to provide custom programming for a telephone system to afford customers with self-service voice. For example, customers calling into a system can be greeted personally by matching their incoming number with your CRM, and then have rendered a custom IVR. Yes, we can integrate with large language models to take a question over the phone, turn that into text, pass that text into a trained Large language model, and then turn its textual response back into voice and play it to the caller. The opportunity and complexity is endless.

Application Example

Imagine if you will, that you're a box shipping company, and you want to provide your customers with the tools to self-service their package enquiries. The customer calls in, is greeted personally, "Good Morning Margaret, you have one package due for delivery, do you want to track it? press 1. Do you want to reschedule it, press 2. To Cancel it, press 3." Assuming the customer presses 1, we render the current status of the package into natural language; "Margaret, our driver Dave has your package and should deliver it to you between 2 and 3pm. If this is ok, press 1, to reschedule press 2"


Hopefully you get the idea here, we can use transcription to really make your company stand out from the rest, at a time where some companies are implementing TERRIBLE chat-bots that do nothing but infuriate customers.


Version 1.009  Copyright © 2024 GEN, its companies and the partnership. All Rights Reserved, E&OE.  ^sales^  0115 933 9000  Privacy Notice