Inquiry Regarding Latency Improvement for Azure Speech Translation

石德平 40 Reputation points
2024-01-09T05:50:18.5466667+00:00

Dear Azure Support Team,

I hope this message finds you well. I am writing to bring to your attention the latency results observed by our client during their testing of the Azure Speech Translation service. The client has conducted tests comparing the latency of Azure Speech Translation with other platforms, and they have observed comparatively higher latencies with the Azure service.

Here are the recorded latency results:

Azure Speech Translation (East Asia):

  • 2.193 seconds
  • 2.240 seconds
  • 2.256 seconds
  • 2.141 seconds
  • 2.198 seconds
  • 2.228 seconds
  • 1.862 seconds
  • 2.227 seconds
  • 2.331 seconds
  • 2.137 seconds

Azure Speech Translation (US East):

  • 2.472 seconds
  • 2.225 seconds
  • 2.099 seconds
  • 2.216 seconds
  • 2.292 seconds
  • 2.302 seconds
  • 2.319 seconds
  • 2.298 seconds
  • 2.310 seconds
  • 2.282 seconds
Client also conducted latency tests for other speech translation platforms for comparison:

**Youdao (SDK Callbacks):**
- From SDK start capturing audio callback to SDK output result callback:
  - 0.924 seconds
  - 0.945 seconds
  - 1.781 seconds
  - 1.203 seconds
  - 1.454 seconds

**iFlyTek (Data Sending):**
- From the start of sending recorded data to the platform to receiving output result:
  - 1.935 seconds
  - 1.372 seconds
  - 1.465 seconds
  - 1.384 seconds
  - 1.339 second

In comparison to other platforms, the client has noticed a significant difference in latency. Therefore, we are reaching out to inquire whether there are any recommended methods or configurations to optimize and reduce the latency of the Azure Speech Translation service.

Azure AI Speech
Azure AI Speech
An Azure service that integrates speech processing into apps and services.
2,074 questions
{count} votes

1 answer

Sort by: Most helpful
  1. dupammi 8,615 Reputation points Microsoft External Staff
    2024-01-09T12:18:39.2866667+00:00

    Hi @石德平 ,

    Thank you for using the Microsoft Q&A forum and for sharing the latency observations.

    To enhance performance, we suggest that you consider leveraging the REST API for Azure Speech Translation. The REST API provides a lightweight and flexible way to integrate speech translation into your applications. Alongside, running a custom model locally in a Docker container can further optimize latency by reducing network latency and processing time. Additionally considering a closer server location, a higher-tier service plan, and enhanced network connectivity can have further improvements.

    I hope this helps! Thank You.

    1 person found this answer helpful.
    0 comments No comments

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.