Introduction
Gary experimented with the ChatGPT voice mode, first released in October 2024. The following recording is raw, unedited, and entirely unscripted.
It’s somewhat concerning how rapidly this technology has evolved. However, it’s important to note that the sound quality is still rough, only a handful of voices are available (with just two British accents), and the regional accents you can ask it to put on are more parody than realistic.
The technology’s potential for use by larger companies is evident - for example, replacing call centre agents and possibly even eliminating the need for IVR menu systems. These companies will likely seek to develop their own 'brand voice' rather than rely on the default voices offered by OpenAI and others. This could present a potential opportunity for professional voiceover artists to ‘lease’ their voice likeness. Such arrangements would likely involve an ongoing retainer fee, with assurances that their voice would only be used for specific purposes and not to train AI models for unrelated projects.
Our motivation for this experiment was to make sure our voiceover artists are fully aware of how this technology is evolving.
Please listen to the audio below. If it doesn’t play, it’s likely due to high traffic from multiple users accessing it at the same time. In that case, please try again later. We start with the British male voice, and then swap to the British female voice, repeating one of the experiments with the female voice.