26th April - Website Update; Platform updates; Reducing bias

We made a bunch of improvements to the platform.

Website update 💯

We made some big changes to our website, improving our design and adding some great new pages. Thanks to the team who worked on this

• https://audiostack.ai/enterprise
• https://audiostack.ai/developers
• https://audiostack.ai/en/blog

Bug fixes and improvements

• Updated file regexes and added decodeURIComponent 
• Fix get colors from org logo error 
• Aesthetic improvements to the sign up flow 
• Add error info on upload Org Profile Picture
• Platform Dashboard Redesign
• Beta Release of SonicSell V6 (enabled for ReleaseRadar only) 
• Audioform/sonicsell endpoint now supports creating more than 1 asset. Exposed through versions to create
• Updated the TTS endpoint to support maxLength per sections. will be used in the sonic sell audioform endpoints.

Docs improvements

• We added some new docs and improved the docs page.
• https://docs.audiostack.ai/docs/writing-a-news-page-audio-summariser
• https://docs.audiostack.ai/docs/using-the-api-to-generate-fully-produced-audio

Reducing bias in our models and systems

One of the challenges with AI models is bias. This is well documented - and has been discussed elsewhere like here
We shipped significant changes, by applying better guard rails and using our proprietary AI guard rails

These changes lead to:

• more gender diversity (before: ~ 1/3 male voices, now: ~50% male/50% female
• less gender mismatches (avoiding e.g. “Hi, I’, m Sarah…” spoken by male voice
• less gender-biased outcomes (e.g. ads on cleaning products have same chance for male/female voice)