Inspiring

avatar creation function

Forum|Forum|20 days ago
January 27, 2026
3 replies
23 views

Hello,
I reported a problem several weeks ago with the “text-to-avatar” AI. I use the same avatar (Tom, French voice) for a series of videos. The issue is that the voice is never the same. It sometimes changes tone from one sentence to the next, or shifts from a very deep voice to a high-pitched one, even to an almost feminine voice. How is it possible that this problem hasn’t been addressed for months? What’s the point of this service if I can’t keep the same voice for the same avatar?

There are other AIs that maintain a consistent voice tone. Honestly, it’s frustrating. Are you going to fix this problem? Thank you for your response.

P

pascal367053547r1qAuthor

Inspiring

the text used :

“Bonjour et bienvenue dans le faike news journal de janvier. On commence par l'allocution du premier ministre Sébastien Lecornu, l'homme de paille du président, qui annonce des mesures pour faire face à la pénurie de logement. Comme vous le savez, les élections municipales arrive ta grand pas. Sébastien Deloggu, le Jean-claude Vandamme de la politique, et occasionnellement député de la France insoumise, se présente à Marseille. En exclusivité il nous dévoile quelques une de ses mesures pour la ville.”

Hajar Toutsi

Community Manager

Thank you for sharing the script. I tested it all at first with both Tom avatars and noticed some sudden changes in voice (including a switch to a female voice) and tone. I achieved slightly better results when I split the script into two separate parts, with the first ending at “ta grand pas.” Unfortunately, there were still some tone inconsistencies.

These inconveniences are expected with a beta feature, and feedback like yours helps us improve the experience before the full release. I’ll escalate this issue to the product team. Please keep us updated if you encounter any other issues, we’d really appreciate hearing about them.

Hajar

P

pascal367053547r1qAuthor

Inspiring

Thank you for your response, but it has been almost a year that this has been in beta 🙂

I would also like to point out that at the beginning of my use of the Tom avatar, there were fewer issues without splitting the text, even though I sometimes had to try several times to obtain a similar voice. Now it is getting worse and worse. Even when splitting the text, it is almost never the same voice. So splitting it sentence by sentence is of no use to me, since there is no consistency from one sentence to the next.

P

pascal367053547r1qAuthor

Inspiring

I recorded this excerpt (video and voice) for you. This is the perfect example of what is not working: the tone of the voice changes in the second part of the text. How do you expect me to make use of this?
Please note that the voice in the second part is the one I used in my first video. Since I am using the same avatar to present a fake news broadcast, I need a minimum level of consistency in the voice. Otherwise, the service is useless.

PJ :

test

Hajar Toutsi

Community Manager

Hi pascal367053547r1q,

Thanks for explaining, and we are sorry it’s been ongoing this long.
To help us reproduce the issue, could you share the exact scripted dialogue you used? Having the text will make it easier to see what’s triggering those sudden tone changes. We would also like to have a look at the generated results to share it with the product team.
Thanks for your patience.

Hajar

Sign up

To post, reply, or follow discussions, please sign in with your Adobe ID.

Sign in to Adobe Community

To post, reply, or follow discussions, please sign in with your Adobe ID.

Scanning file for viruses.

This file cannot be downloaded