AWS transcribe optimal solution for 3500 calls per day.

Hi, AWS Transcribe allows you to queue up to 10'000 jobs while 100 execute in parallel. So, you can submit your 3'500 requests all at once if you receive them in one shot (or incrementally if not). AWS Transcribe will manage their parallel execution: https://docs.aws.amazon.com/transcribe/latest/dg/job-queueing.html

Note: you can increase the defaults if needed. Place a quota increase request in that case

Using job queueing, you can submit more transcription job requests than 
can be concurrently processed. Without job queueing, once you reach the 
quota of allowed concurrent requests, you must wait until one or more 
requests are completed before submitting a new request.

Job queueing is optional for transcription job requests. Post-call analytics 
requests have job queueing enabled automatically.

If you enable job queueing, Amazon Transcribe creates a queue that contains 
all requests that exceed your limit. As soon as a request is completed, a new 
request is pulled from your queue and processed. Queued requests are processed
 in a FIFO (first in, first out) order.

You can add up to 10,000 jobs to your queue. If you exceed this limit, you get a LimitExceededConcurrentJobException error. To maintain optimal performance, 
Amazon Transcribe only uses up to 90 percent of your quota (a bandwidth ratio 
of 0.9) to process queued jobs. Note that these are default values that can be 
increased upon request.