Forum Discussion
AI accents changing - Rise 360
I have seen that there have been other discussions about the accents of the AI Text to Voice changing. My colleague and I have both had similar experiences.
I have been using the Addison 2.0 Australian Audiobook & Podcast voice in the majority of my courses, and had not had an issue with it until a week or two ago. The accent changed from Australian to New Zealand. Before I checked the discussions here, I tried to see if the accent would change if I just kept clicking the Generate Speech button - and it made very little change.
I then checked the discussion board, and tried the suggested fixes, first entering an Aussie term at the beginning of the audio "G'Day" - which did make the audio revert to the Australian accent, but as soon as I deleted G'day, as it was not appropriate for my course, it reverted to the New Zealand accent. I then tried to change this to the Multilingual v2 and the voice was even worse.
After no luck with the accent for this block, I went to enter the next audio block further down in the course. The voice was perfect, and all of the other audio blocks in this course and others have been fine.
Today, I returned to the course, to see if the accent would change if I tried to re-generate the audio. No luck. Due to not having the issue in the rest of the course, I tried to create a new audio block in the same position. Accent was still the same. I tried to create a new audio block under the following text block and still no luck. In the end, I added a new audio block at the end of the lesson. Addison has two versions, so I tried to generate the audio with the Addison - Australian Narration and Conversational, and this generated the Audio with an American accent. I then changed it back to Addison 2.0 - and finally have the Australian accent.
My colleague has also had issues with the following Australian voices having American accents: Emily, Jess and Clara. When she has changed the settings to the Multilingual V2, the volume of the audio decreases.
While we are loving the AI tools, this issue when it is occurring is taking up way too much time to investigate and where possible have it work as it should. Leaving it there, or removing the audio entirely will impact the learner expereince.
Is there any planned rectification works for these issues?
10 Replies
It's probably not much help, but I'm having a similar issue - although for me, it is Max, who started out nicely with a New Zealand accent but then quickly changed to an American accent, while occasionally slipping in to a northern UK accent. I have given up trying to get him to speak like a Kiwi. At least regenerating the audio removed the UK accent. I may go and try Addison 2.0!
Hi SuePage,
Thank you for adding your experience with Max. I’m sorry the voice changed between New Zealand, American, and UK accents and that regenerating the audio didn’t restore a consistent New Zealand accent. I understand how frustrating that can be, especially when you’ve selected a voice expecting a consistent Kiwi accent.
If you haven't already tried it, Multilingual v2 is the model we recommend trying for accented voices. You can find instructions for switching models in this article.
I'd also like to better understand what this has meant for your workflow. How has the inconsistent accent affected the way you’re creating or updating narration for your course? Since you’ve been unable to get a consistent Kiwi accent from Max, what are you doing for that narration instead?
Hi JMManalili , I did try Multilingual 2, but I still couldn't get a consistent New Zealand accent. My solution has been to generate the audio in Articulate 360, using the same voice, but adding a "Kia ora, bro" prompt at the beginning, then trimming this out of the audio clip before exporting it, and then importing it into Rise. (This was a suggestion by another member). It's a bit more long-winded, but Max is much better for it. (The ability to do the same directly in Rise would be amazing.) I have also been using Ella, and that voice seems to hold a much more stable New Zealand accent in Rise, without the need for a prompt. I think I will play around with more of the voices and find a few favourites to use. Even with needing to manipulate the files, it's still quicker than having to record the audio myself!
Hi EmilijaPether,
Thank you for sharing such a detailed account of what you and your colleague experienced. I understand how disruptive it is when a voice’s accent changes between generations, especially when you need consistent narration throughout a course.
Our current recommendation is to use the Multilingual v2 model for the most reliable results with voices that have accents. Since you’ve already tried this step and it didn’t produce an acceptable result with Addison, I’d love to better understand the impact this is having on your workflow.
I don’t have any updates to share about changes to this behavior at this time, but the additional context you’ve provided helps us better understand how the inconsistent accents are affecting your experience. How often are you encountering these accent changes when generating narration? When this happens, what does it typically mean for your course development process beyond the troubleshooting time you’ve already spent?
Thank you again for taking the time to explain everything you tested.
I am currently working on a suite of courses and, to ensure consistency across the learning experience, I am using the same AI voice, Addison 2.0, throughout. As I mentioned in my original post, I had been achieving very good results with this voice until recently.
The issue first became apparent when the voice-over for a single block in the course was unexpectedly generated with a New Zealand accent. As mentioned in my original post, I had to use a process of trial and error to eventually generate the audio with the desired Australian accent. However, after successfully achieving this and later needing to make updates to the same block, regenerating the voice-over caused it to revert back to a New Zealand accent. As a result, I was forced to switch back to the original Addison voice, which unfortunately does not provide the same level of quality as the Addison 2.0 version.
The accent change was particularly noticeable because the block contained several words that are pronounced differently in Australian and New Zealand English. This made the inconsistency quite obvious and would have been distracting for learners, resulting in an unacceptable learning experience if left unchanged.
You asked how this affects our workflow. My colleague and I typically work to tight delivery timelines, so these issues have a direct impact on our ability to meet deadlines. Omitting the voice-over is not a viable option, as it reduces accessibility and support for auditory learners. Instead, we need to spend additional time testing various workarounds that have been suggested in previous posts, hoping that one of them resolves the issue. The rework is adding additional time to complete tasks that the AI functionality should be assisting us with.
Another challenge is inconsistent pronunciation. For example, the name of our CMS may be pronounced differently across blocks, despite using the same voice. Sometimes regenerating the audio once will resolve the issue, but in other cases multiple regenerations are required before achieving a consistent result. This adds further time and unpredictability to the development process.
One enhancement that would significantly improve this tool would be the ability to provide pronunciation guidance or feedback when generating audio. For example, allowing users to enter instructions such as "pronounce XXXX as YYYY" would provide greater control over the output and help ensure consistency across a project.
My colleague will also be leaving some feedback relating to the issues she has experienced, and the impacts it has had.
Hi EmilijaPether,
Thank you for coming back with these additional details. The context around tight delivery timelines, the extra rework involved when an accent changes, and the importance of keeping narration consistent for learners gives us a much clearer picture of how this affects your course development.
I also appreciate you raising pronunciation control for specific terms. Your CMS example highlights another challenge with maintaining consistent narration when a particular word or name needs to be pronounced a certain way.
Thank you again for taking the time to give us such a detailed look at your workflow.
Hello,
I am Emilija’s colleague and have been experiencing the same issues with the Australian voices. I primarily use the Australian voice Clara - Australian Professional Female with warmth & Trust, and have also tested Jess - Australian middle-aged women who is genuine and friendly and Emily - Australian female, however all three have intermittently switched to either an American or UK accent.
I have attempted the recommended workarounds, including adding prompts such as “G’day mate” and “Australian accent”, as well as using the Advanced Settings to switch to Multilingual v2. While this occasionally corrects the accent issue, it is not a consistent solution and often results in a noticeable reduction in volume.
Consistency of voice is critical to the learning experience. Having different accents appear within the same course can be distracting for learners and as customers of this product, this impacts the overall quality of the content and our experience.
We appreciate the suggested workarounds; however, I would think expectation is that the voices selected are as described and should perform consistently without ongoing intervention.
I have also noticed that when filtering for Australian voices, some non-Australian accents appear in the results. It may be worthwhile reviewing the catalogue and filtering logic, as well as checking all options are useable, as there are a number of samples I am unable to preview (they do not play). Apologies, I have not collated results of theses.Thank you for looking into a fix for us users and training content creators.
Hi RobynJarvis,
I appreciate you taking the time to test several Australian voices and share what you found. It’s helpful to know that you’re seeing the accent changes across different voices, and that switching models hasn’t consistently resolved them.
The other behavior you noticed while browsing the voices, including samples that wouldn’t play and voices that didn’t sound Australian to you, is helpful context as well. It gives us a better understanding of what you’re running into when trying to create consistent narration.
I'm also experiencing changing accent issues while using the AI audio generation tool in Rise. I have been using Australian Emily in courses throughout early 2026 and had no issues until around August 2026, when I tried to use her voice in a new course. I tried a few different things such as switching to 2.0 etc. but that only confused her more.
I couldn't spend any more time figuring out a workaround, so left her half UK/US/Aussie accent until a viable solution becomes available. I'm now back working on that project, and would really appreciate a workable solution before the project launches in November 2026. Has anyone heard when its likely to be rectified?Hi LeahMayes,
I appreciate the additional context around what you’re experiencing and the impact it’s having on your project. I understand how important consistent narration is, especially with your November launch approaching.
I don’t have a timeline or an update to share on changes to this behavior. I know you’ve already spent considerable time trying different voices and workarounds just to keep your project moving. Hearing how this affects your project gives us valuable context around the experience.
