The sentence It is the 100% correct pronunciation of 'mischievous' you just heard in the video is grammatical, but it is ambiguous, and it is unidiomatic.
It is grammatical because it agrees with the rules of how English noun phrases can be constructed. The pattern of the 100% correct pronunciation of 'mischievous' can be analysed as DETERMINER + ADJECTIVE PHRASE + HEAD, which is a rather frequent pattern in English. Struturally, the noun phrase at hand may be analysed like so:
So, there's nothing wrong with the grammar of your sentence.
But your sentence is ambiguous because you just heard in the video could either be a relative clause or a subordinate that clause.
If it is a relative clause that refers to the nominal pronunciation of 'mischievous', then the subject it of the sentence is refers to one particular recording of the word 'mischievous', namely the one that was just heard in the video and which happens to be also a 100% percent correct pronunciation (regardless such a thing exists in the first place – see the comments to the OP). As such, there is no particular focus on the subject. Structurally, the relative clause clause is you just heard ___ in
the video, with ___ indicating the gap from which the 100% correct
pronunciation of 'mischievous' was extracted. It is integrated into the noun phrase [NP the 100% correct pronunciation of 'mischievous' [that] you just heard in the video] as a post-head modifier.
If it is a subordinate that clause, then the whole sentence is an it-cleft that places strong focus on the noun phrase _the 100% percent correct pronunciation of 'mischievous', as in It is the 100% correct pronunciation of 'mischievous' [that] you just heard in the video – NOT the pronunciation that is only 50 percent correct. Structurally, there is a huge difference to the relative clause, as the that clause is subordinated to the matrix clause it is the 100% correct pronunciation, and not a post-head modifier as in the relative clause interpretation.
The expression is unidiomatic because (American) English speakers appear to prefer the indefinite article in noun phrases that contain 100 percent + ADJ as an attributive modifier of the head. This is supported by numbers from the Corpus of Contemporary English (COCA). This can be tested using the search string
a|the 100 percent ADJ NOUN, which searches for combinations of 100 percent and any adjective which is followed by a noun preceded by either a or the.
The definite article is found in only 12 tokens, at least five of which are questionable because it's not clear whether 100 percent modifies the following adjective or acts as a modifier of the nominal that consists of the adjective and the head noun. Here are the remaining seven cases from COCA that should have a structure analogous to the one in your sentence:
After London, the [100 percent recyclable] sculpture made the rounds on a tour
And the [100 percent federal] reimbursement of debris cleanup […]
The President is proposing an extension of the [100 percent expensing] provision that he signed into law
articles attacking their marriage, such as the [100 percent misguided] prediction that Odom 's departure…
Sex education programs should mention that the [100 percent conclusive] way to avoid unwanted pregnancy […]
"Why?" "The [100 Percent American] Act. For genetic security, they say. […]"
We came up with the middle-class tax cut plus the [100 percent tax-deductible] tuition... Excuse me.
Tokens in which the construction 100 percent + ADJ + NOUN is preceded by the indefinite article a are much more frequent in COCA. The total number of hits is 75, and while I didn't go through all of them, I'd estimate that at least 60 of them are cases that contain an Adjective Phrase [100 percent + ADJ] as in your data. Here are the first seven confirmed instances:
the SAFEST bet is to shoot for a [100 percent healthy] weight and stay there?
If we get a [100 percent accurate] vote count in Florida, then every state's […]
A [100 percent UV-protected] polypropylene cover will be available […]
A [100 percent recycled] pile fabric made from plastic soda bottles by […]
Gambineri added that untreated rabies can cause nearly a [100 percent fatal] illness in humans, and that the viruses […]
of securing our airports . Although it seems clear that a [100 percent secure] airport is years away , it is important to realize
[…] diabetes patients who did not use insulin had a [100 percent increased] risk of developing cancer […]
The log-likelihood test is one of the established ways of determining whether the frequency difference between two corpus searches is statistically significant. I used the Lancaster LL calculator for that purpose. Indeed, assuming a frequency of 7 for the definite article, the lower estimate 60 for the indefinite article, and a corpus size of 1 billion for COCA, the difference is indeed statistically significant (LL=48.02, p<0.001). The effect size estimators such as the Bayes factor show that this is a very strong effect (Bayes=26.06).
Based on this, my prediction would be that if you did an acceptability rating task for your original sentence (which includes the definite article), and if you then repeated the task with the indefinite article, your speakers would find the second variant notably more acceptable.