While ASR-ers don’t actually perform DBT or similar nearly as often as they demand it of others
If a DBT is done, you have to count each test, no matter the number of participants, as a single data point. Because the test results may be biased. So, the number of tests required to be statistically significant goes well past the majority of tests ever reported. Make the sample of tests large enough to be statistically significant and you are at the man on the street level of participants, not trained listeners. And that is how you get Bose level of audio quality.