Hello! When audio is uploaded into Architect, it is transcoded into a format that can be played on an edge device. Currently that format is 8-bit sample size, 8khz sample rate, single channel (mono), µ-law encoding. So, as it stands right now, that is intended behavior. Higher quality audio recordings may not be playable on the edge, and so that's why Architect does this transcoding; to make sure all uploaded files can be played on the edge. We may revisit this to see if we can switch to higher quality audio formats, but at the moment, unfortunately the audio format cannot be specified on audio uploads in Architect.