added
31st May 2024 - Better sound quality; JS SDK updates; Performance improvements
6 months ago by Peadar Coyle
Bumper improvements π
We made a bumper set of improvements this week π―
Better sound quality with media files
- Β We enhanced the mastering sidechain compression for music using media as key signal, allowing to combine File Manager with TTS audio in the mix in a leveled way. This enables better sound quality in more situations π
API Bug Fixes
- πWe made some improvements to the Audio Evaluation Service bugfixes
- Better Errror handling improvements when Files were Not found β οΈ
Platform fixes and bug fixes
- π Rebranding of Audio Playground to Workshop
- πWe fixed a bug to always show org switcher
- πWe fixed the "you are not authorised" flashing when page loads
- π fixed re-rendering in asset card and a bug with the play button
- πfixed indexed Db error when db is manually deleted
- π We added new icons to the platform add-new-icons-to-platform
- π We fixed some bugs in the library
- π fixed ghost card bug
Platform and SDK features and improvements
- π― SDK for Suite endpoints, long polling and error handling. You can check it out now in the Python SDK
- π― added lazy mode to asset cards to enhance the customer experience
- π― Added pagination to library so now it will load faster π
- π fixed conditional for play button icon in asset card
- π We updated default sound design to
soiree_a_illange
across all workflows
JS SDK v1.1.11
We improved the JS SDK and launched a new version https://www.npmjs.com/package/@aflr/audiostack
-
Integration with STS (speech to speech) Endpoints:
-
https://v2.api.audio/speech/sts/voicesΒ (GET to get a list of speech to speech voices available)
-
https://v2.api.audio/speech/stsΒ (POST to start an sts pipeline)
-
https://v2.api.audio/speech/sts (GET to get the status of the pipeline launched)
Performance improvements π€
Speed improvements for SonicSell, which was causing some errors
- Reducing deployment package size of ms-script. ~600ms faster.
- Calls to recommend mood/gender/tone in parallel, saves 4-5 seconds.
- We reduced cold starts in our infrastructure also yielding up to 1.8 seconds saving.
This should make your experience much faster π€