Startup Profile
PHONT

Introduce yourselves

Tell us in three sentences who you are

PHONT combines artificial intelligence and design to bring subtitles to life. With the help of a specially developed AI, we recognize language characteristics such as emotions in spoken language and represent them with empirically developed design. The result is emotional subtitles that are an immersive and accessible way to experience language visually on social media and streaming platforms.

What challenge does your startup address and how did the idea for it come about?

The idea arose from the observation that type has so far only done limited justice to its potential for speech reproduction. Co-founder Frederik had tackled this issue in his type design master’s thesis and designed lively subtitles. The challenge was to communicate and validate the product vision and now to integrate it seamlessly into the relevant platforms.

Can you briefly explain how your solution operates and to what extent AI supports this solution?

PHONT analyses voice recordings with AI algorithms to recognize emotional states, voice and speech characteristics, as well as sound events. This information is then displayed in real time in the subtitles using visual parameters to make the linguistic content visible to the viewer.

How does PHONT differ from conventional subtitles and how does the integration of emotion recognition help to improve the viewing experience?

In contrast to conventional subtitles, which only reproduce the spoken text content, PHONT also conveys the rest of the information on the audio track. This enhances the consumption of all audiovisual media through greater emotional depth, attention-grabbing power and accessibility. Social media consumers who like to watch their content on mute and deaf people in particular benefit from this. Users can decide for themselves what their subtitles should look like.

What steps are you taking to integrate AI for recognizing emotions in spoken language into your service and how are you adapting this to different languages?

PHONT integrates AI by using machine learning and natural language processing to recognize emotions in spoken language. Language-specific models are later trained to adapt to different languages, taking into account cultural and linguistic differences to ensure accurate emotion recognition in different language contexts. In this way, we avoid biases of any kind.

Let’s take a look into the future: Where could the project be in five years’ time?

In five years, PHONT could be an established standard in the media industry, used by numerous content creators, streaming services and educational platforms worldwide to provide the new quality standard of subtitles.

Interview with PHONT