Geert-Jan Strengholt (VPRO Medialab)
In the Netherlands, approximately 1.5 million people are hard of hearing or deaf. As such, this audience has limited or no access to the growing volume of public and commercial podcasts. This means that they miss out on the valuable information and the rich world of stories uniquely offered by podcasts.
During a VPRO Medialab hackathon on augmented podcasts, inspired by BBC’s Ian Forrester and his notion of adaptive podcasts, we asked ourselves how we could make podcasts accessible for a deaf or hearing-impaired audience while retaining their unique quality in format, dialogue and atmosphere in storytelling. As our prime motive consisted of making existing podcasts or podcasts in production accessible, rather than developing a whole new format for this audience, we decided to build upon an existing script or transcript. Based on the prototype of the hackathon, provisionally entitled DoofPod, we conducted interviews with users from various target audiences.
Prototype player
The current prototype of TypeCast ‘plays’ a textual transcript of the original podcast, by presenting – or typecasting – lines of text. This transcript is ‘tagged’ in order to recognize voices/speakers, voiceovers, settings/contexts and ambience. The flow of text emulates the flow of sound and the development of the story experience.
The transcript is augmented by closed captions, specifically written to capture atmospherics and moods generally conveyed by sound designs. Here the explorations of artist Christine Sun Kim were a source of inspiration. Special care is taken to avoid the use of words that refer to sound qualities that a deaf audience would not know or understand. By way of abstract visuals the soundstage is set, presenting and denoting separate speakers within the podcast, but also the presence of background sounds.
This rich script is then played by a mobile first code-player, dynamically calling on and including visual and functional assets… without the original audio track.
User testing
In late 2022, the TypeCast player went through user tests within various user groups, ranging from a ‘sudden or late deaf’ audience (who still have Dutch as their native language versus sign language) to an audience with limited or severe loss of hearing. We aimed high by featuring one full episode of the award-winning Dutch/ Belgian podcast Bob, by the audio collective SCHIK, a signature example of VPRO’s rich and immersive productions.
What became immediately apparent was that the preferences of these test users pre-empted a one-size-fits-all solution. Those who were hard of hearing preferred having the audio track augmented by text, whereas people born deaf were overwhelmed by the amount of text and visuals.
Generally, the runtime of 40 minutes was experienced as too long, as TypeCast is much more ‘lean forward’ than an audio podcast. Currently VPRO MediaLab is exploring an object-based audio approach in order to allow the audience to customize TypeCast to their needs. An updated version for public release is expected in June 2023.
This article was first published in issue 55 of tech-i magazine.