Audio Audit

Pod Bible is a UK-based print magazine, podcast and website. The print magazine publishes a new issue every two months which mainly contains interviews with industry experts and recommendations of podcasts to listen to.

We have a half-page ad in their latest issue #021 (June 2022) and package also gets us host-read ads on their future podcast episodes and an opportunity to write an article with their digital editor for their website.

Hopefully this is helping to increase brand awareness but it's too early to attribute any increase in traffic.

You can read more here: https://audioaudit.io/articles/news/pod-bible-magazine-issue-021

It's official — Audio Audit is now launched to the general public! We put together a press release (https://audioaudit.io/articles/news/press-release-public-launch) and now we're reaching out to journalists today to hopefully help spread the word. Glad that we could announce this in time for The Podcast Show in London on Wednesday.

We just turned on our subscription pricing for Audio Audit!

We've tried to strike a balance between making the business sustainable and it being simple and free to get started. We also go through our payment flow, choice of payment/subscription provider (Paddle) and thoughts on the future. You can read more in the post. The full launch is hopefully very close now.

Analysing our customer’s audio files is quite intensive, requiring quite a lot of CPU and RAM resources to generate a report within a reasonable period of time.

With our new architecture, more worker machines get added when more work needs doing and then they are terminated (switched off when not in use). We use Kubernetes as our container orchestrator running on Google Cloud using their Autopilot feature. Traditionally you would have to fix the size of your Kubernetes cluster upfront but Autopilot gives us the flexibility and simplicity we need as they deal with adding more node machines to the cluster as we increase the number of pod replicas in our specification.

Read more:

Thanks to our new server-side backend code, we can store more audio files than ever before. We can robustly handle many more customers at one time and provide a faster service. It took several days of solid work to make sure all parts were transitioned over but we think it will pay dividends way into the future.

The takeaway is that we’re now better able to handle customers and generate reports for them. We invest in quality, modern software and think you should know. Here’s to more customers, more reports, and more storage!

Read more:

The registration form was enabled on the site so people could create an account. I chose to approve accounts manually as I didn't want to overload the server as the speech recognition phase is quite CPU intensive.

I would work on parallelization and scaling up the infrastructure while beta testing was in progress.

Bought the domain name audioaudit.io and launched a simple landing page to start with. This started the process of becoming known to the search engines. The site began to highlight the main features of the tool.

I was able to reuse the descriptions of audio checks used in the reports and package them into individual public pages to help with content marketing.

I planned out a good way of storing the various types of data extracted from audio data and how they would collectively form a report for the user.

React was used to gradually build components that would make up the various parts of the dashboard, report, player etc.

To prove that I could detect the types of problems I was hearing I collected small samples of audio with each type of flaw from podcasts and elsewhere.

I wrote Python functions for each type of measurement and check whilst writing the test suite - test driven development!

I was listening to lots of podcasts and audiobooks while insulating and decorating a bedroom.

I noticed there were occasionally repeated sections left in where the voice artist wanted to re-record. Also I was constantly having to turn the volume up and down between podcast episodes, which was tricky when your hands are covered in glue and paint.

I did some research and found there were audio standards for things like loudness (which not many podcasts were following). I figured I could detect restarted sentences by first performing speech recognition on the audio.