Research using presigned URLs for uploading audio to S3

[citomcclure]

https://docs.aws.amazon.com/AmazonS3/latest/userguide/PresignedUrlUploadObject.html

Using presigned URLs will have several benefits, including:

• Faster performance since we are note sending audio through HTTPs, decoding, parsing, validating audio format, and then writing to temp before finally uploading to S3
• Current max Lambda invocation payload size is 6MB
  • AWS Lambda documentation here
  • and since audio is getting Base64 encoded, it's more like 4.2-4.4MB according to this
• Will make updating notes with additional transcription (see scope change here Voice notes #26 ) implementation much easier since we will be able to send data in HTTP now (like dateCreated field)
• typically considered the "correct" way to do it and is prone to less mistakes
• will make it easier for researching Research optimizing media file format for transcriptions #33

[citomcclure]

If unable to implement presigned URLs, easier (but smaller) optimization is to stream audio directly to S3 instead of creating and saving a file to execution environment (i.e., temp)