I was an early employee at Whisper so I got to see many of the challenges in this space first hand. I can't speak to the specifics of Whisper leaving the market, but I can say generally that bringing a new piece of medical hardware to market is very hard for a startup to pull off.
In the early days you need to bring together a group of people with a diverse set of skills --- mechanical engineers, electrical engineers, firmware engineers, audio engineers, regulatory specialists, people with experience working with factories, along with ML engineers in our case. You need a lot more people to bring a product to market than if you're building another SaaS company. (And, of course, all your code is running on a chip that has to be small enough to fit behind their ear.) Then hardware startups also have a problem that they tend to have slower growth and lower margins than successful software startups. So later stage VCs are more inclined to invest in software startups.
Hearing aids are also unique in that they require a lot of tuning to get the audio to sound as natural as possible. There are dozens of algorithms that process different features of the audio, things like wind noise reduction and feedback cancellation, and some of them can be a bit finicky. Companies like Oticon or Widex have hundreds of engineers who have been tuning these algorithms for decades. (Another advantage the big guys have is that they can make custom silicon So if you want a top of the line hearing aid the bar is very high. Then until recently there were big barriers to entry on the sales side --- you weren't selling just to the customer, you were also selling to the audiologist, and many audiologists have incentives to (mostly) sell a single brand. This is less of a barrier now with the recent changes to allow hearing aids to be sold over-the-counter.
I do hope to see another startup give it a go because I still think there's room for disruption --- the big companies move a lot slower than a startup can, and the ODML space is developing very, very quickly. But it's not a project for the faint of heart.