ACE - Lipsync Issues with Omniverse A2F

[psneha]

Hi,

We are attempting to automate the lip sync pipeline using Nvidia ACE Audio2Face microservices.

We have implemented the animation pipeline workflow as suggested in NVIDIA Animation Pipeline with Omniverse Renderer ( Docker & Omniverse Animation Pipeline Workflow — ACE documentation) . The setup works with default-avatar-scene:1.0.0, but the lip sync quality is not great - Default avatar

However, when using our custom avatar, we followed all the required steps detailed in Custom Avatar Creation**. Despite this, the resulting custom avatar does not perform any lip sync when run through the same pipeline as above - Edited - custom avatar rendered with ACE pipeline

We are using audio generated using Azure TTS, 16 kHz sampling rate - [Audio_azure_tts](https ://drive.google.com/file/d/1GORk8gg9cwabCxBNJ9niu1nXKhuPO5E-/view?usp=sharing). We also tried with Nvidia’s TTS, the results were the same.

Would you be able to guide us on what could possibly be wrong?

[ttripathi]

Please review this troubleshooting guide to synchronize audio and video using the “livestream.audioDelay” parameter to the appropriate value: Troubleshooting — ACE documentation

[Richard3D]

I am sorry but we do not really support ACE or A2F here on this forum. That would be on our Nvidia Developer Discord channel.

[psneha]

Hi @ttripathi , The documentation link seems to be invalid. Can you reattach a valid one

[Richard3D]

Here is the correct link
NVIDIA ACE — ACE Overview

We’ve Moved NVIDIA A2F Support! We have new places to better help you get the support you need.

1. Developers can submit tickets through the NVIDIA AI Enterprise program NVIDIA Enterprise Customer Support
2. Developers can discuss ACE through our NVIDIA Developer Discord Server NVIDIA Developer (no GeForce)

[ttripathi]

Here you go: Troubleshooting — Animation Pipeline

[Richard3D]

Thanks