I'm using Azure text to speech service. I have a lot of small ssml files for which I generate the audio, then combine them using FFMPEG.
I wrote a script to generate all the small audio files. If the script makes 3 concurrent requests, then after about a total of 20 requests, some of the other requests will complete, but the file will be empty. Checking the result object, I see Throttled due to too many requests websocket error code: 1007.
If I remove any concurrency from the script and add a 1-second wait between the end of the last file and the next request, it works.
I've checked the quotas and limit on the Text to speech API and it mentions a maximum of 20 concurrent requests (which is far from the 3 I send).
Any idea about what I might be doing wrong (I'm using the javascript SDK).