Audio clips should generaly be digitally mastered before you import them to Spark AR Studio.
This helps to ensure that none of the imported sounds dominate the experience.
When creating your ambient audio assets, consider:
If you create a smaller, seamlessly looping ambient sound asset, you'll need to edit the file in a digital audio workstation, such as Pro Tools, and make the edit cut on the zero crossing at the sample level. This edit is essential to creating a seamless loop, so make sure you cut it so that the wavelength at the end of the file and the wavelength at the beginning of the file flow continuously into each other, without a fade applied to either end.
Below are two examples:
When mixing ambient beds, we recommend that you record audio from a couple of mobile devices and import those files to mix against in your digital audio workstation. An effective ambient bed, much like environmental sounds or a score in a film, should sit well below the production audio (or a person's voice in a capture video), so creating some dummy audio captures will help you achieve the proper levels. You'll be able to make adjustments in AR Studio (only in lowering dB levels, currently), so testing on a mobile device and mixing levels before implementation will save you a lot of time on the implementation stage.
With scripted audio assets, consider:
You can use a one shot - a trigger that plays only once all the way through - to trigger a scripted sound.
If you plan to couple ambient beds with scripted effects, make sure to mix both assets together and against the production audio references that you capture from your test mobile device. You'll be able to adjust scripted audio assets in Spark AR Studio just as you can adjust ambient beds, but it's best to get your levels set before importing.