Noise assessments measure people not places
Here's an important point to remember in any noise assessment - the Regulations could not care less about any noise levels from equipment or machinery which cause no noise exposures to people and have no potential to do so. The Noise Regs are entirely concerned with noise levels experienced by your employees so the noise levels of machines are not the focus of it but merely a part of it.
What do we mean by that? Well, if you have Billy Bob who works on a punch press for some of his day, and when he's not there is using a chop saw and a lathe, then the noise assessment for him is a combination of all three noise levels and how long he is exposed to each, not just the noise levels of the individual tools.
The logical extension of these gets us to the Noise Map. This is the thing where you have a drawing or plan of the workplace and have noise levels marked on it for each tool or piece of equipment. While this can be useful as a tool for a very high level overview of noise patterns on the site, especially a complex site where it lets you get a good picture of the patterns of the noise levels, it is NOT a noise assessment by itself.
So, a noise assessment should be job or individual based. This doesn't mean you have to measure every single person, especially in a large factory where you may have 20 people doing exactly the same job in the same area with the same equipment. In this case one or two measurements will give you a reasonable estimate of the noise levels experienced by all. Be careful though to make sure nobody is at increased risk by working next to a reflective wall or something, but otherwise it is a reasonable way to go.
Exemptions to measuring the person
With the best will in the world, sometimes measuring the person or job function just doesn't work and the best example of this is the humble joinery workshop. The problem here can be that what an individual does on one day may have no relevance whatsoever to what they did the day before or will do the day after so calculating a daily noise exposure for the single day in question is fairly useless.
In these situations, I switch to measuring the noise for individual tools or equipment rather than the person as otherwise the results are fairly meaningless.