Monks, Microsoft and the Machine Learning Dividend

Transcription software saved my (business) life. Here's how the latest tools can help you.



Even in the days of multi-media content, I still spend a lot of time interviewing clients and customers before turning these conversations into headlines, captions and copy. In most cases I’ll record the call, although more often I get an audio file via Zoom or Google Meet after the event. It’s a time-consuming business. Some weeks I’ll rack up to four or five hours of recordings.


Mind you, that’s nothing compared with when I ran a content team which produced dozens of case studies every month for tech clients. We sometimes put six hours on the clock every day. In those days, about 10 years ago, the writers took notes and kept a dictaphone recording as back up.


There wasn’t time or budget to get a transcription, so after every call the writer would go through their notes, filling in gaps from memory, and even typing up the most useful quotations into a rough draft. By today’s measure they were medieval monks dutifully copying religious and medical texts onto vellum pages.


Words, words, words

In recent years I’ve used a few free or low-cost transcription services, some of them human, some of them automated. But they all have their weaknesses, especially when the call quality is low, or the conversation heavily accented.


My transcription hack is to play a tricky interview through my headphones while reading it aloud to Microsoft Word or Google Docs, both of which have built in dictation software. Sometimes it helps to slow down the recording by 25% to capture every syllable. It’s a clunky method, but it works.


Step forward then Microsoft, which is adding a transcription feature to the upcoming version of Office 365. To the delight of business writers everywhere, you can now upload an audio file and in a matter of seconds the full transcription appears in a side bar.


It looks really simple to use. Once the text is there, you can jump around the time stamps and copy segments directly into a Word document. Ideal for isolating the best quotations and assembling them into a rough draft.


Everyday AI

It’s also the perfect example of how AI is becoming a part of our everyday software. Microsoft have made a big deal out of the AI features in Office 365 and in my opinion they are now the leading software company when it comes to embedding the technology into their brand.


They’re also on the leader board when it comes to cloud-based AI. These are services that businesses use to analyze data and train mission-critical software at a fraction of the cost of on-premise tools. Amazon and Google offer something similar of course.


As I’ve written before, one of the most overlooked advantages of AI is that it is universally applicable. Any business that ignores its own data and the new family of cloud-commoditized machine learning tools is like a farmer cutting down corn with a scythe while a gleaming new combine harvester sits idle in the barn.


Another misapprehension is that AI is only for digital transformation or once-in-a-generation projects. It’s far more versatile and flexible than that. Many of the most useful AI applications are incremental improvements that address an urgent end-user need. The spike in demand for video conferencing during 2020 resulted in quick fixes that addressed audio drop outs and unflattering camera angles.


As for me, I’ll be using the Microsoft transcription feature straight out of the box. It’s going to save time, offer better value for clients, and free me up to get on with left-behind admin and marketing projects. Sign me up. I’m all set to go. As soon as I get my Office 365 upgrade, of course.

Feel the connection?

© 2020 Peter Springett.