Real-world data is the competitive edge in AI. HumanSignal is a human data partner for companies building AI models and products. Our customers ship better AI, faster, because we partner with their researchers from real-world data creation to annotation to delivery.
Heidi Health – UK Accent Speech Study for Clinicians (£10 per approved set of recordings, ~15 min)
United Kingdom
October 16, 2026
Job description
Heidi Health, the clinical AI scribe used by clinicians in the UK, Australia and the US, is building speech recognition that works for every accent. To get there, Heidi needs short recordings from UK-based clinicians who speak with specific regional and community accents.
What you'll do -Record yourself reading two short passages aloud (one clinical, one non-medical) from a quiet space, using your phone or laptop microphone. -Total time: around 15 minutes. -Before any paid recording, and as part of the screening process, Heidi listens to a short unpaid sample to confirm your accent and audio quality. If it isn't a match for this study, we'll tell you before you start. The sample is only used for that check. It will not be used to train speech recognition or any other model.
Who we're looking for UK-registered healthcare professionals (GMC, NMC or equivalent) who live in the UK, read English aloud fluently and naturally speak with one of these accents day to day: -African (Nigerian, Ghanaian, East African) -British-African -British-Caribbean -East Asian -Lancashire -Liverpool and Merseyside (Scouse) -Middle Eastern -North East (Geordie, Mackem, Teesside) -Received Pronunciation (RP), Standard Southern, Estuary or South East English -Welsh -West Midlands (Brummie, Black Country)
We're asking how you speak, not where you or your family come from. We don't need your nationality or ethnicity. Please refer to our accent guide to help identify the group that best describes your speech: https://drive.google.com/file/d/1svkV64RpCfpRRFf1WxhjdJgZOhCegHqM/view?usp=sharing
Pay £10 ($13.35) for each completed set of recordings (around 15 minutes in total). You'll be paid once Heidi confirms your recordings are complete and clear.
How your recordings are used Heidi will use your paid recordings, together with the accent you select, to train and test its speech recognition. Because an accent can indicate ethnic origin, we ask for your explicit consent during screening.
Places are limited per accent group, so apply early.
Requirements
- UK-registered healthcare professional (GMC, NMC or equivalent), practising now or recently qualified
- Lives in the UK
- Reads English aloud fluently
- Naturally speaks with one of the eleven listed accents day to day
- Has a quiet space and a device with a microphone
- Agrees to Heidi reviewing a short unpaid sample before paid work (the sample is not used for training)
- Consents to Heidi using the paid recordings and selected accent to train and test speech recognition
Apply now
Opens agent.askethos.com in a new tab. If they ask how you found the role, mention RemoWork.
We are looking for AI Voice Recording Contributors to join a French (Cameroon) voice evaluation project and help support the development of the next generation of AI technologies.
iMerit seeks Hindi Language speakers to join as Gen AI Audio Trainer Analysts in our global remote team. Analysts will perform audio annotations to deliver high-quality datasets essential for training and refining cutting-edge Generative AI models.
Innodata (Nasdaq: INOD) is a global data engineering company. We believe that data and Artificial Intelligence (AI) are inextricably linked.