A repo to document some code related to Azure Speech to text's batch transcription features using the REST APIs. Source files from here, the here, and here.
- The general flow of operations is captured in the Makefile.
- Note that the audio files to be processed through a batch transcription request need to honor the following formats. There is a command in the Makefile which illustrates tooling to aid conversion (at least on a Macbook).
- To run any of the commands (on a Unix/Linux system), at the terminal, you would run
make <command>
. - For example, once you've authenticated to your Azure environment through the Azure CLI, to deploy the infrastructure (a speech resource and a storage account to
host the audio files), you can run
make infra
, which creates a resource group, provisions the infrastructure, etc. This also takes the files insource-audio
and uploads them as a batch. - Once that runs successfully, you can then run
make create-transcript
to initiate a batch transcription. - Then, run
make get-status
to check on the status of the job request. - Run
make get-files
andmake get-file-transcripts
(in that order) to download the transcriptions.