Warning: this content is older than 365 days. It may be out of date and no longer relevant.

Mind Readings: Adobe Podcast Review

Join Christopher Penn as he reviews Adobe’s Project Shasta, a new AI-powered tool for cleaning up audio. In this episode, Christopher puts the software to the test by recording in various challenging environments, including a moving car with high road noise and in a parked car while wearing a p100 mask. Tune in to see how well Project Shasta performs and how it uses artificial intelligence to improve audio quality.

Mind Readings: Adobe Podcast Review

Can’t see anything? Watch it on YouTube here.

Listen to the audio here:

Download the MP3 audio here.

Machine-Generated Transcript

What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for watching the video.

Christopher Penn: In today’s episode, we’re going to do a bit of a bake off and review of Adobe Podcast.

If you’re not familiar, Adobe rolled out this product, which has been in beta for a while.

It’s called Project Shasta, and now it’s in production and people can try it out.

I imagine eventually it will find its way into Adobe Audition or something similar.

In a nutshell, it’s a pretty cool product.

It uses artificial intelligence to clean up sound, and we’ll talk about the type of AI it uses, because it’s not what most people think.

But first, let’s talk about the audio clips we’ll be listening to.

The audio clips in this episode are going to be audio only, because a number of them were recorded in a moving vehicle.

For safety reasons, there’s no video because I was operating the vehicle.

The four samples we’ll be looking at were recorded in a parked car, just on a phone; in a moving vehicle with the ventilation system on high; in a moving vehicle with high road noise; and in a parked car while wearing a P100 mask.

This last scenario is not ideal for having conversations, as the mask is great for stopping bad things in the air from getting in your lungs, but not for having clear conversations.
Let’s listen to each of the samples and then discuss what the software does.
[Audio samples played]
Okay, it’s pretty clear that Adobe Podcast does an incredible job with some really terrible audio.

All four samples were very short because they were difficult to listen to.

We did hear some distortion, of course, because the recording scenarios were poor.

But the really interesting thing is that last sample with the P100 mask.

It did a very credible job of trying to reconstruct my voice.

That’s an important part of the product – it’s not just doing the usual leveling tools you find in audio software, it’s actually reconstructing the voice.

It has a trained library of known voices, so it can compare the distorted audio to the known voices in its library and try to reconstruct the original audio as closely as possible.
Overall, I’m really impressed with Adobe Podcast.

It’s not perfect, but it does a fantastic job given the terrible audio it’s working with.

If you’re interested in checking it out, you can try it out now that it’s in production.


You might also enjoy:


Want to read more like this from Christopher Penn? Get updates here:

subscribe to my newsletter here


AI for Marketers Book
Take my Generative AI for Marketers course!

Analytics for Marketers Discussion Group
Join my Analytics for Marketers Slack Group!