… a nice opportunity to show off your machine learning capabilities – specifically with transcription help for my next book.
I tend to interview lots of executives for my books. For each of my last 6 books I have tried some speech-to-text technology – various versions of Nuance Dragon, Siri etc. with very little impact. I keep going back to human transcription services.
With ML having matured so much – and voice interfaces starting to show up in so many applications – would love to have the transcription be automated this time. Over the next 3 months I will likely have 100 audio files to transcribe.
Many of these files will be captured from a speakerphone on a Zoom recorder, and I will be able to upload those MP4 file. So, there will be some degradation due to the call quality. Also many of the speakers will have German, other European, Indian and other accents and will use tech jargon. But the conversations will be in English.
Amazon, you have the inside track. We use your KDP platform to publish the Kindle version of books. Having said that, the first use of AWS Transcribe did not produce very good results.
Google you also have an inside track. Most of these MP4 files are stored in Drive. I got excited when I saw the GCP Speech to Text conversion, but my conversations are longer than a minute. Can you help with longer conversions?
IBM, Microsoft, Oracle – same offer to try out Watson, Azure and Oracle ML services on my project.
Look forward to your support!
(Cross-posted @ Deal Architect)