Speech to Text

Pedro Rodrigues 0 Reputation points
2023-03-30T11:02:01.08+00:00

Good morning,

I am exploring your Speech-To-Text with the aim of creating a service that makes calls, asks a question, analyzes the response, and records it. For each call, an instance is created to isolate the transcription. The service streams audio to Azure using audioInputStream.Write(data.Data, data.Data.Length), starts StartContinuousRecognitionAsync, and waits for translations to record the responses. It works perfectly for a single call, but the problem arises when there are multiple calls. How many calls can I transcribe in real-time at the same time? Is there a limit to creating instances of the SpeechRecognizer class? Can I only use one?

Best regards,

Pedro Rodrigues

Azure AI Speech
Azure AI Speech
An Azure service that integrates speech processing into apps and services.
1,383 questions
0 comments

2 answers

Sort by: Most helpful
Most helpful Newest Oldest
  1. Dillon Silzer 54,286 Reputation points
    2023-03-30T15:49:59.1033333+00:00

    Hi Pedro,

    If you are using the Free version, then you are limited to 1 concurrent request limit (call). If you are using the Standard plan, you can make up to 100 concurrent calls:

    Online transcription and speech translation

    https://learn.microsoft.com/en-us/azure/cognitive-services/speech-service/speech-services-quotas-and-limits#online-transcription-and-speech-translation

    If this is helpful please accept answer.

  2. Pedro Rodrigues 0 Reputation points
    2023-03-30T17:17:42.2466667+00:00

    The problem that seems to be happening is that even though I am sending the audio stream of the calls separately, when I have more than one call and therefore make more than one stream, I stop receiving responses as soon as I hang up the calls, and only when I have a single call again everything starts working. My current hypothesis, which seems more credible to me, is that when I stream the audio, even though I am sending it in two different instances, for the Azure service it is like it's just one stream, and I stop receiving responses because the audio no longer makes sense.