Overdub Pronunciation Tips

Descript's Overdub technology allows users to create speech audio by simply typing text, either with our custom Overdub voice technology, or by choosing one of the many Stock Voices from our library. While Overdub does a great job most of the time, getting perfect pronunciation can be a bit of an art form and may require several iterations.

This guide provides suggestions for how to approach the process, including working with common words that may have several pronunciations, as well as tips for getting less common words to resolve as intended.


Getting Started

Correcting the pronunciation of a word is typically a three-step process:

  1. Adjusting the spelling of the word
  2. Converting the Overdub clip to audio
  3. Correcting the spelling on the transcript



Let's see how this might work in a real-world scenario with the phrase "The rapidly diminishing rainwater presents an opportunity for the crocodiles."



In this case, Overdub renders the word "presents" with the accent on the first syllable as if it were a "gift" (present) rather than the second syllable as intended. This is common when distinguishing between a noun form of a word vs the same word in verb form.


We need to find a way to adjust the spelling to arrive at a desirable pronunciation. The first attempt of "pruh-zents" gets a little closer, but still has a bit of a halting tone.


Updating the spelling to "prezents" seems to be the best version for our case.


Correct the Spelling

Once you're satisfied with the result of the pronunciation, hover your mouse over the Overdub clip in the Timeline Editor and click the Convert to audio button.



Now that the audio has been converted, you may use the Correct Text feature (letter 'E' shortcut) to set the word back to the correct spelling.




You may run across proper names with unique spellings or pronunciations. Try either using the more common spelling, or experimenting with alternate spellings that may be more phonetically similar.



The name "Braden" is typically rendered as "Brad-en" with the short 'a' sound.


By adjusting the spelling of the word to "Brayden" you can achieve the long 'a' version of the vowel instead.



Initialisms are an abbreviation consisting of the initial letter of multiple words, that are then pronounced individually - such as DMV (Department of Motor Vehicles). Initialisms can pose a particular challenge for Overdubs as the letters are occasionally rendered as an acronym (such as NASA), rather than the individual letters. In these situations you'll achieve best results when spelling out each letter as if it were its own word.



The word IRA will likely be pronounced "ear-uh" by default.


Updating the spelling to "i-array" will achieve a much more accurate result.


Noun / Verb Stress

In the English language, it is common for words that have both a noun and verb form to accent a different syllable. Although not true for every case, these stresses typically follow these guidelines:

  • Noun - Stress the first syllable
  • Verb - Stress the second syllable



One common way to address the issue of stressing different syllables is to use a hyphen to break up words into more distinct sections. For example, the noun "combine" (as in a combine harvester used in farming) is rendered as the verb "to combine".


You can use a hyphen to break this up into "kom-bine" and achieve the correct pronunciation.