Automated transcription service screenshot

Automated transcription service

At the University of Nottingham, researchers and students spend over £250k a year on transcription services, the current experience is costly, slow and unsafe. This project was part of a multi-million-pound initiative to produce an automated transcription service that would help improve the quality and efficiency of research conducted at the University.

I worked within a multi-disciplinary project team made up of a business analyst, architect, tester, project manager, product owner and an external development agency.

It was my responsibility to advocate for the end-users, applying a design thinking methodology to the project and involving users throughout the project from user research to testing.

Timeframe: 6 months

Responsibility: UX Lead

EMPATHISE

Competitor analysis

To understand the area of transcription services I conducted a competitor analysis of three manual transcription services, most used by researchers at the Univerisity. The focus of the competitor analysis was user journeys, features, content and interactions. The competitor analysis was presented back to the project team, creating a shared understanding of user expectations and journeys whilst using transcription services.

Key findings

  • Security is paramount.
  • Using an online transcription service is complex, requiring several email responses.
  • Transcription services are expensive and costs increase based on the complexity of the language and desired turnaround time.
  • The webform should be a maximum of three steps

EMPATHISE

User interviews

Researchers are traditionally hard to reach, often time-poor and working off-campus. I managed to recruit ten participants for the interviews from a range of subject areas and varying levels of experience with using transcription services.

The interviews were conducted over a month at a location and time that suited the participants. The interviews were then analysed and the findings were shared with the project team using personas. The research helped the group focus and consider real users’ needs during the project.

Key findings

  • Each research project is different, from the number of people in a project, budget and location.
  • Accuracy of the service and limited budgets means the service is unsuitable for some researchers.
  • Automated transcription could allow some researchers with big data sets to explore further.
  • Three key user groups emerged from the interviews and they have been reflected in personas.
examplar persona

One of the three personas derived from the user interviews.

DEFINE

What's in scope?

This project is one of several workstreams that make up the research initiative. The transcription project has dependencies on the self-provisioning portal which is another workstream, without it, researchers won't be able to log-in to consume the transcription service.

As a project team, we had a collaborative session, mapping the entire research initiative. Creating an overview of the landscape, it's dependencies and risks. From that, we defined the scope and started to build requirements. I visualised the overview of the transcription service and detailed some of the requirements to share with the project board.

System overview on the whiteboard

As a team, we mapped the flow and sope of the system.

System overview

I formalised the system overview to share with the project board.

IDEATE

Sketches

I chose to sketch my ideas for the design and flow of the service, it was a quick and lean approach that allowed me to generate lots of ideas in a short space of time.

The sketches were informed by the research conducted earlier in the project. I took a clean and minimalist approach to the design, creating a simple UI which was easy to use and in alignment with other transcription services.

Sketches

Low-fidelity sketches of the upload process.

PROTOTYPE

How will a researcher use the service?

Midway through the project, the board were having difficulty envisioning how the transcription service would work, and they were hesitant to continue funding the project.

To help communicate to, and get the buy-in from, the project board I created a high-fidelity prototype. I worked closely with the project team and drew inspiration from the research conducted earlier in the project.

Key findings

  • Testing the prototype helped highlight flaws and allowed us to rectify them before starting development.
  • The design was much quicker, reducing the effort in the sprint so we could concentrate on functionality and the journey.
  • Funding was secured on the back of the prototype, as it helped communicate the ambition to the project board
Prototype screenshot

The high-fidelity prototype of the transcription service.

TEST

Usability testing

I ran two sets of usability testing, one at the beginning of the development and one nearing the end. The first set of usability testing involved four participants, a mixture of remote and in-person sessions.

The second set of usability testing was conducted before the last sprint, that gave ample time to make any changes before the end of the project. The findings from the testing were quickly shared with the team on our Microsoft Teams site.

Key findings

  • Cross-browser compatibility prevented some participants with older versions of IE from uploading files.
  • Logging-out of the transcription service would also log users out of Office 365, this would be an issue for some participants.
  • The navigation and design were easy to use.

OUTCOME

Faster, safer and cheaper transcription service

The delivery of the automated transcription service means that researchers can transcribe their work much faster, hours instead of days. Their work will remain safe, no sending memory sticks or sending files to external companies. Costs are much lower than manual transcription, researchers can transcribe larger data sets, improving the validity of their work.