Given the recent shift in scholarly attention from the quantity to the quality of media accessibility (MA), this paper aims to provide an account of how the assessment of quality in MA may be negotiated across researchers, regulators, companies and users. Taking as an example the use of the NER model to assess live subtitling quality, it focuses on what each party needs to do and compromise on, so that a consensus can be reached and quality in MA can be assessed. It covers firstly the role played by researchers, who are normally expected to develop rigorous models of assessment while, at the same time, ensuring that they are transferable to the industry. The focus is then placed on media regulators, which must often make a choice between adopting a soft or a hard approach. The former tackles quality only superficially and may have little effect on professional practice or on the viewers’ experience. The latter requires investing time and resources to conduct research, but it also results in tangible improvements in quality. As for companies, they must be willing to accept external reviews and even to modify their training, but they can also benefit from the improved performance of their staff. Finally, in order to have a voice in this area, the users may be expected to engage with the method and the results of the quality assessment. This can enable them to have more informed and realistic demands that can lead to effective changes and, ultimately, to better access.

鉴于最近学术界关注的焦点从数量转移到媒体可访问性（MA）的质量，本文旨在提供一种方法，说明如何在研究人员，监管机构，公司和用户之间就MA的质量评估进行谈判。以使用NER模型评估实时字幕质量为例，该模型着重于各方需要做和做出的妥协，以便可以达成共识并可以评估MA的质量。首先，它涵盖了研究人员所扮演的角色，通常期望他们开发严格的评估模型，同时确保将其移植到行业中。然后将重点放在媒体调节器上，媒体调节器通常必须在采用软性或硬性方法之间做出选择。前者仅从表面上解决质量问题，可能对专业实践或观众的体验影响很小。后者需要投入时间和资源来进行研究，但同时也可以显着提高质量。对于公司，他们必须愿意接受外部审查，甚至愿意修改培训，但他们也可以从员工绩效的提高中受益。最后，为了在这个领域有发言权，可以期望用户参与质量评估的方法和结果。这可以使他们拥有更加明智和切合实际的需求，从而可以进行有效的更改，并最终改善访问范围。但这也带来了质量上的明显改善。对于公司，他们必须愿意接受外部审查，甚至愿意修改培训，但他们也可以从员工绩效的提高中受益。最后，为了在这个领域有发言权，可以期望用户参与质量评估的方法和结果。这可以使他们拥有更明智的，更切合实际的需求，从而可以进行有效的更改，并最终改善访问范围。但这也带来了质量上的明显改善。对于公司，他们必须愿意接受外部审查，甚至愿意修改培训，但他们也可以从员工绩效的提高中受益。最后，为了在这个领域有发言权，可以期望用户参与质量评估的方法和结果。这可以使他们拥有更加明智和切合实际的需求，从而可以进行有效的更改，并最终改善访问范围。