Chokehold analysis
So I have done some analysis to figure out exactly why the end of phrase /d/ sounds in chokehold during the verses are perceived as /d/ sounds instead of /t/ even though they are unvoiced. So to start here is the /d/ in “made” and “flood.”
The vertical line in the middle top of the /d/ in the spectrogram is the stop burst. Note that it lines up with higher amplitude at the wavelength that correlates with the amount of air being released. In this case it’s atypical for the placement in the word. English speakers tend to not release as much air after it stops that are not at the beginning of words and are not voiced. But the other thing happening here is that voicing stops at the end of the vowel. This is not typical for a /d/ this matches a /t/.
When we measure the time between voice offset and a burst of the stop it is about 478 milliseconds which is a really long time for speech. Once again this should be a /t/. Acoustically this sound out of context looks more like a beginning of word /t/ than an end of word /d/. Which then creates a number of questions about why we perceive a /d/.
I looked at the /d/ in flood to double check my work. It's very similar. In this case we can see a reduction in amplitude where Vessel begins to devoice and then a similar drop when voicing is fully faded out. Notably even without voicing formants are present both before and after the stop to create an unvoiced /ə/ sound. Once again acoustically this should be a /t/.
A possible explanation of the perceptual /d/ sound can be found by looking at the /t/ in “accident”
The /t/ we have that is perceived as a /t/ has an even higher amplitude meaning even more air released or aspiration. The most surprising is that the /n/ sound that the /t/ follows is also not voiced. These two help explain what is happening. This /t/ sound is extreme in its distinctions between /d/ and /t/. And this distinction level causes the other /t/ sounds to by comparison be more ambiguous between /t/ and /d/ even when acoustically they are not. Added with semantic context that a /d/ sound would make more sense in those places we can perceive a /d/ in those places.
So is Vessel doing this on purpose or not? I have no idea. The aspiration is a choice. The devoicing is a choice. That is purposeful. The further differentiation between the /t/ in accident. I am not sure if that is something he has full awareness of. But what I am sure of is that it’s fascinating. Thank you for coming to my ted talk :)
