Director Mode works best for English text.
WHY? The speech is generated in a slightly different way from normal clips in the studio, meaning that it can sometimes mispronounce text that isn’t in English, regardless of which voice is applied.
Specify the accent if you want non-American.
WHY? Director Mode produces results in an American accent by default, so you need to specify in your prompt if you want a different accent such as British or Aussie, regardless of which voice is applied. You will get the best result by being very precise about the accent, leaving no room for interpretation.
PROMPT: “Speak with a London millennial accent.”
Use on shorter text segments for more precise editing.
WHY? If you apply Director Mode to a long segment with multiple sentences, you have less control over how the individual parts of it sound. You’ll also have to regenerate the whole thing with each take. This is likely to take longer and use up more credits. Aim for 10-15 lines maximum.
Be precise with your prompts. Don’t leave room for interpretation.
WHY? If your instructions aren’t precise enough, the AI will make its own interpretation of what you want, which can lead to hallucinations or inconsistency between clips. If you find that the transition between clips that have Director Mode applied is obvious and unsmooth, you should make sure you are using similar instructions for each clip.
PROMPT: "Speak clearly, with a moderate pace, and a confident, upbeat tone, like that of a radio host.”
Be instructive about the pace.
WHY? If you don’t specify what pacing should be used, the AI will make its own interpretation based on whatever context it has.
PROMPT: “Start speaking very slowly, then speed up gradually with each word, leaving a 2-second pause just before the final word: [insert final word].”
Remember that Director Mode remembers context within the same chat.
WHY? When iterating on a clip in the same Director Mode chat, it will remember all instructions given. This means that if you previously told it to speak with a British accent, then, in the following message, tell it to speak with an Indian accent, it will try to combine the two instructions. In this case, if you want to clear all context, you should exit Director Mode and then open it again to start a new chat.
Don’t close Director Mode until you’ve finished iterating.
WHY? Once you’ve closed Director Mode or clicked Approve Take, you won’t be able to access the chat history or previous takes, so make sure you iterate in one go if you want to keep the context of all instructions. If you Approve Take and then decide you want to edit the same clip with Director Mode again, you will have to repeat any necessary previous instructions regarding accent, pacing, etc.