I find that I'm often correcting the transcription for certain words like "and" where my speech pattern doesn't enunciate the end "d". I know this is user error, but it really doesn't sound good when I pronounce a hard "d" at the end. I'm curious if the transcription tool can be further enhanced by using AI to optimize it's performance by being trained on my specific voice. I trained a speaker AI on my voice, but it doesn't seem to further get trained on my voice as it has more information. I'd love to give this program full access to further optimize on my vocal style (which would communicate with the transcription function optimizing both functions).

I love Descript! I'm a voice actor using Descript to help me edit out all of my stumbles and repeated sentences when I do narrations. It's easier to edit as a document than manually listen for all of the stumbles in my DAW.

Something I thought would be super cool would be an option to upload a text document to sort of "train" the transcription as to what I am supposed to be saying. Then, theoretically, it would be able to find and highlight the sections in the transcription where the words I'm speaking don't match the provided text - thus making it easier to find and kill all of my stumbles.

I'd love the ability to run a content library through Descript, and have it learn from internal guidelines and writing style.

For example, I want Descript to apply usage guidelines from this site (https://contentdesign.intuit.com/word-list/) to everything it transcribes.

An example would automatically changing "okay" to "OK", or making sure words are hyphenated when applicable.