Transcribe Unlimited Audio Minutes for Free(ish)

Say Goodbye to Word Count and Time Limits with this Simple Trick:

Have you wanted to transcribe your meeting recordings, a video playback, or podcast you recorded but found that all the “free” online tools limit you to a word count or meeting length that is only suitable for a test or two, not real production work? Well, I discovered a great little method to get effectively free, unlimited transcription using tools you likely already have available to you.

Update! May 24, 2023:

Thanks to the stellar review by William Gallagher on the pros and cons of Whisper Transcription available for Mac users (free and Pro versions), there’s an even easier and more free way to transcribe without length or size limitations than the kludgy way I describe below. So, if you’re running MacOs, I’d suggest downloading Whisper and moving forward with ease… and if you’re on Windows, well, I’ll still stand by the below methods.

Now, if you’re not keen on downloading Whisper, I’m going to walk you through the steps to use the transcription service provided by Word for the Web to get effectively free and unlimited audio to text transcription. Since you likely have access to this through your employer’s Office365 subscription, or your own personal Microsoft Office account, this “hack” is effectively free to you. It’s also important to note here that this feature is currently only available on the Word for the web service.

We’re going to leverage that weird little “Dictate” button that most of us likely haven’t even touched before….

The transcribe feature contained within converts speech to a text transcript with each speaker individually separated. After your conversation, interview, or meeting, you can revisit parts of the recording by playing back the timestamped audio and edit the transcription to make corrections. You can save the full transcript as a Word document or insert snippets of it into existing documents.

You can transcribe speech in two ways: record directly in Word, or upload an audio file. Since the most common use case is likely from a recording, I’ll only be discussing that method:

Upload an audio file

1. Sign into your Microsoft 365 account using your favourite browser like Chrome, Firefox, Edge, or Arc (which has quickly become my new personal browser of choice).
2. Open an empty Word doc.
3. Click the down arrow just below the Dictate icon
4. In the Transcribe pane, select your appropriate language and click the Upload audio button.
5. Choose an .wav, .mp4, .m4a, or .mp3 audio file
6. The file will begin to be transcribed and the progress displayed in the open pane. Make sure to keep the Transcribe pane open while the transcription is happening.
7. Once complete, you will see the full transcription displayed with time codes and speaker identification.
8. At the bottom of the pane, you will be prompted with a choice to Add to document (with a drop arrow) or New transcription.
9. Select the Add to document drop arrow to see a selection for Just text, With speakers, With timestamps, or With speakers and timestamps. Make your choice based on your use case.
10. Your document is now populated with the generated transcript for you to continue your work.

And it’s as simple as that. Rinse and repeat as often as you desire… for free(ish)!

Dictating in Word is similar to the above so I’ll spare you the steps, but, if you absolutely need explicit steps you can refer to the MS Knowledge base here.

The transcribed text generated by Word for the Web’s transcription service could now be used as input data for generative AI models. By feeding the AI tool the transcribed text it could, for example, generate blog posts with summaries and steps to complement your Youtube videos or podcasts of the same topic. This could be useful for various other applications as well, such as creating chatbots or just delivering summaries and action items from your meetings without having to take copious notes, giving you more time to thoughtfully engage in the meetings you attend. But be ever vigilant as it is crucial to carefully review and edit the transcribed text to ensure the accuracy of the input data, as I’ve found (like most things) it isn’t perfect.

Only users with Microsoft 365 subscriptions can transcribe unlimited minutes of uploaded audio using the transcribe feature, for now… As an Enterprise subscriber, you may need to request an Unlock for unlimited upload minutes. Since this is currently what Microsoft is calling a Preview Experience, the unlimited capabilities may not last forever.

So, for now at least, I’d recommend using it to your best benefit personally, but not tooling your systems around the assumption it will always be so open or free.

---

 

With regards to security and privacy of your content: There are ongoing discussions around the privacy, and lack thereof, with regards to all AI tools. Some companies have outright banned their employees from using services like ChatGPT because of instances where corporate proprietary information could be accidentally used in public answers generated by the tool. \

Like with any service provided, users must be proactive and vigilant around their use and agreement to the terms of service. Likewise, it is smart to never put anything into these tool that you wouldn’t want publicly known, much like how we should all view what we post to the internet in general.

To help, I asked ChatGPT about this transcription service’s privacy, and this was the reply:

Acdntlpoet: How secure is the transcribe function within Microsoft Office?

ChatGPT: The Transcribe function within Microsoft Office is reasonably secure, but there are some potential security risks to keep in mind.

First, when using the Transcribe function, your audio files are uploaded to Microsoft’s servers for processing. This means that there is a risk that your audio files could be intercepted or accessed by unauthorized third parties during transmission.

Second, Microsoft stores the transcribed text generated by the Transcribe function on its servers. This could potentially create a risk if the transcribed text contains sensitive information.

However, Microsoft has implemented various security measures to protect users’ data. For example, all data sent between the client and Microsoft’s servers is encrypted using Transport Layer Security (TLS) protocol, and Microsoft also uses multi-factor authentication and other security measures to protect user accounts.

In addition, Microsoft provides users with controls over their data. Users can choose to delete their audio files and transcriptions at any time, and they can also choose to opt out of certain types of data processing.

Overall, while there are some potential security risks associated with using the Transcribe function within Microsoft Office, the risk can be minimized by taking appropriate security precautions, such as using a strong password, enabling two-factor authentication, and avoiding transcribing sensitive information.

 

So, before implementing this solution, remember: caveat emptor.

Image by rawpixel.com