Resolving Pause Issues in Azure TTS with zh-CN-YunxiNeural Voice Pack

Question

How can I improve the pause issues when using the zh-CN-YunxiNeural voice pack in Azure Text-to-Speech (TTS)?

PS - Based on common issues that we have seen from customers and other sources, we are posting these questions to help the Azure community.

Answer 1

Greetings!

When using the Azure TTS service with the zh-CN-YunxiNeural voice pack, you may encounter unwanted pauses in the speech output. To address this issue, you can try adjusting the prunebreak parameter in the Speech Synthesis Markup Language (SSML). This parameter helps control the pauses in the synthesized speech. Alternatively, you can consider using the zh-CN-YunxiLTSNeural voice pack, which might offer better performance regarding pause handling.

To adjust the prunebreak parameter, you can include it in your SSML input like this:

<speak>

    <voice name="zh-CN-YunxiNeural">

        <prosody prunebreak="medium">Your text here</prosody>

    </voice>

</speak>

Experiment with different values for prunebreak to find the optimal setting for your needs.

If these steps do not resolve the issue, further assistance from Azure support may be required.

Resources:

• Azure Text-to-Speech Documentation
• SSML Reference for Azure Cognitive Services

Hope this helps. If you have any follow-up questions, please let me know. I would be happy to help.

Please do not forget to "up-vote" wherever the information provided helps you, as this can be beneficial to other community members.