Is AI Transcription False Economy? | Virtu@dmin
Artificial intelligence has developed over the years and more recently with accuracy improvements many businesses are using these apps and websites. But, is AI transcription false economy?
As a manual transcriptionist with many, many years of experience, I really would like to address this question, “Is AI transcription false economy?” for the many entrepreneurs, journalists, small businesses and anyone that uses transcription services.
Of course, as a transcriber when a robot comes along that is now automating my job I am going to see it as competition, but is it, is AI transcription false economy for the business or person paying for that service? This is what is in my mind as to why it may not be so…
What is audio transcription?
Firstly, before we get into the in’s and out’s of artificial intelligence and its role in audio transcription, in short, let’s describe what audio transcription is. Audio transcription is the process of converting recorded dialogue into a written, readable, comprehensible document known as a transcript. An interview transcription example is shown in the left-hand-side image. It identifies when the speaker changes, would an AI transcription bot do that?
There are a few factors that can impede the transcription process and these apply whether it is automated or manual human-powered conversion of the audio file into text. These factors can determine the accuracy of your transcript, even human-powered transcription! It’s all about the quality of the recording, the number of speakers, the topic, and accents of speakers. Do the participants speak over each other, known as cross-talking when transcribing.
These are the challenges for us manual transcribers that can slow down the process of converting the spoken word into text, what would artificial intelligence programs or apps produce?
The manual human-powered transcription process
For a manual transcriber once we have downloaded the audio file for transcription it is then uploaded to specialist transcription software, for example, Express Scribe. Software like Express Scribe allows us, transcribers, to play the audio file with our foot by pressing on a foot pedal. This means we do not have to take our hands off the keyboard and can stop, play, and rewind the audio with our feet. This speeds up the transcription process immensely.
The quality of the audio can delay the process if the quality isn’t that great. The reason for this is that a lot more replay of the audio file will be required to try and attain a better understanding of the content for improved accuracy of the transcript. If there is static, background noise or the audio is just low in recording volume, this can cause delays. Sometimes, the professional transcription software can provide certain processes that can help clear the audio, but this isn’t guaranteed to work, to assist the transcriptionist. If we have to do this for an audio file before the actual transcription process, what would an AI app do?
As a manual transcriber, I have had projects, jobs in the past that required me to correct artificial intelligence-created transcripts. On occasions, this is for real, it was quicker to transcribe the audio from scratch than to try and correct the mistakes of the AI-produced transcript!
Is AI-transcription false economy?
So, above, I’ve gone into a bit of detail about the challenges that a manual transcriber can face when attempting to transcribe an audio file. The standard as a guide for the transcription, because people speak differently at different rates, (and the aforementioned issues) is that for one hour of recorded audio it takes four to five hours to transcribe it into text (transcript), usually a Word document.
Therefore, when you use speech recognition software or a service that uses automatic transcription software that it is cheaper to transcribe your audio file than manual human transcription services, quick to turn it around and deliver the transcript back to you from your audio, then what? Well, you’re not going to just trust it and use it, are you? I hope not! No, you then have to invest more time in checking it, doing that yourself or by delegating it to someone that can do that task for you. More of your time, or more money if you don’t do it yourself.
Don’t get me wrong, depending on the intended use for the transcript, for some this purpose artificial intelligence transcription may suit perfectly well. Just to receive the text from the recorded audio file in one document. But, will it have the correct grammar? Will it be intelligently paragraphed? Or, are these things that you or someone else has to undertake afterwards?
See where I’m going with this?
Artificial intelligence is great to get the recorded spoken word into text, a document with the content, but even then it probably won’t be that accurate due to different factors already mentioned previously. So, it’s cheap to start with, upload your file, receive the transcript back in minutes or hours but then there is still more work to do. Still, more time needed to get it to the usable, readable, comprehensible standard you need and want.
If you have someone available that can do that, great, but then you are taking them from the task they are doing to undertake this one. Will that have an impact on your business? If you haven’t, that means either doing it yourself or more expense and outsource it to someone that can proofread it and format it correctly, possibly a manual transcriber!
It all depends on the intended end use of your transcript. If you’re going to use it to communicate with others details of a meeting, design details for example, then you need it to be as accurate and word-for-word as possible. Thinking about the uses of your transcript after your investment this is what other purposes it can have, see our post “ 9 Ways to Use Your Audio Transcript”
Conclusion
Artificial intelligence is a great tool for journalists, writers and such like. Those that need just the text so they can copy, paste and edit the document for readers to read it their way.
But, for those who want to use it with no additional processes involved, other than merely converting their recorded audio file into a word-for-word account of what was said, then artificial intelligence is not the way to go. Although it is initially more affordable and quick, the time checking afterwards and editing, the time that involves and time is money then it is simply a false economy.
For an accurate transcript, with correct grammar and punctuation, you need to hire a manual transcription service. In the long run, the extra expense, although it isn’t that much more with us at Virtuadmin compared to other companies. What you will receive is a ready-to-use transcript for whatever you want.
Our rates for our manual transcription services start at just 60p per recorded audio/video/podcast minute. Included in the fee is proofreading, editing and you can have the document in a format you would like.
When you receive your transcript you will then be able to use it, distribute it, or anything else with it from the outset. No checking through the document against the audio. Just receive your transcript, upload it to your computer and it’s money well spent with time-saved.
Contact us directly at we would be please to speak with you and provide you with a competitive quote for our transcription services. We are specialist in construction and medical but have undertaken a lot of general transcription too over the years.
You can alternatively see our profile on People per Hour at pph.me/virtuadmin. We have great feedback on many of our services, and that includes copy typing, construction survey transcription, podcast transcription and show notes, medical transcription and more. We offer bespoke services should you require it and many happy clients have left their feedback about our services. You can therefore hire our transcription services with confidence. Please visit our testimonials page to read more.
Originally published at http://virtuadmin.uk on August 6, 2020.
