At Interesante, we’re enhancing our character with interactive features so it can engage directly with the audience. Interactivity can be achieved through a voice and/or the implementation of body movement. After research and market analysis, we determined that giving our character a voice was more important than body movement. This decision was based on the observation that most characters created with AI don’t use movement in their content. These characters post pictures on their socials or videos using only images with an AI voice narrating their experiences, suggesting that the voice is a more crucial element for connecting with the audience.
For us, it was essential for the character to have a voice, as it adds a dimension to their personality and establishes a deeper connection with our audience. Moreover, the voice will allow us to use it for storytelling, podcast creation, interviews, and more. The voice can convey unique emotions, tone, and style that complements the visual appearance of the character. Although not strictly necessary for all characters, having a voice can make the experience more immersive and memorable for the audience. For example, trends on social networks show that videos with narration and a voice interacting with the public are highly engaging and successful today.
Although we considered the possibility of giving the character body movement, we realized that this task is complicated and requires skills we currently lack. Additionally, the rapid evolution of text-to-video models, like Sora, led us to believe that soon, tools will be available to facilitate the creation of videos with movement. Therefore, we decided to focus on developing a voice for our character, as this option was more accessible and would aid our content strategy effectively. In this article, we will detail our exploration of creating a voice with text-to-speech AI models, discussing the challenges faced with current products and how our character will contribute to making our content more engaging and interactive for our audience.
The Quest for the Human Touch
Text-to-speech technology has advanced significantly recently, becoming a crucial tool in various applications. Although mechanical devices from centuries ago tried to imitate the human voice, today, thanks to the rapid advancement of artificial intelligence (AI), we have models that can replicate human speech surprisingly realistically. Technology companies worldwide focus on developing AI tools, and text-to-speech technology is one of the most sophisticated and researched areas.
Numerous technology companies now offer text-to-speech models in various applications, from narrating audiobooks to providing real-time navigation instructions in GPS apps. This technology is useful in daily life and allows users to experiment and create new means of communication in an accessible way.
In our case, we wanted to endow our character with a unique voice close to human speech to establish a deeper connection with our audience. After exploring various options, we decided to use Eleven Labs, a voice synthesis and text-to-speech software that offers exceptional quality, price, and ease of use.
Initially, we tested Eleven Labs’ default voices, which are ideal for creative projects and narrations, but they did not replicate human speech exactly. The main limitation of using text-to-speech technology is the robotic feeling that the voice conveys. To eliminate this effect, we opted for the human voice cloning feature. With this feature, we uploaded the voice of Ale, one of the project’s co-authors, to the Eleven Labs platform and used their AI models to replicate her voice with 90% similarity. With Ale’s cloned voice, we conducted experiments to determine how our character would interact with the audience. We believe this voice will make our content more engaging and interactive and also help us establish a distinctive and memorable identity on our digital platforms.
AI avatar’s voice using Eleven Labs’ cloned voice feature
Useful Tips When Attempting to Create Interactivity
- Prioritize interactivity to differentiate your character and make it more attractive to your audience.
- Giving your character a voice is crucial for producing high-quality content, even if it is only used as a voiceover for a video.
- Consider using Eleven Labs for its excellent quality and accessibility. Its voice cloning feature opens opportunities to create diverse content.
- Regarding movement, it is advisable to wait, as current tools are still in development and may not significantly contribute to enhancing interactivity.
- Carefully evaluate tools like Runway and PikaLabs, as they might require a lot of time and not deliver the expected benefits in interacting with your AI avatar.
Lessons We Learned During This Process
- Enhance your character’s voice using text-to-speech tools like Eleven Labs, known for its quality and affordable price.
- Prioritize voice cloning if possible, as it gives your character a more realistic and expressive voice.
- Establish a workflow for incorporating interactivity into videos, although this remains challenging. If you do not have an easily replicable method, consider postponing movement integration into your character for the moment.
