Is it ok to use very short audio ？ #229
-
|
They come from a game and the typical length is 0.5s to 1.5s, they mostly consist a few words or no word with emotional sound like crying. |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 2 replies
-
|
One second of the speaking part is sufficient. |
Beta Was this translation helpful? Give feedback.
-
What if it's just sobbing or laughing ? or just one word |
Beta Was this translation helpful? Give feedback.
-
|
Sorry for the late response, but too much diversity in the way speakers speak can have a negative impact. You cannot know what will happen unless you try it. |
Beta Was this translation helpful? Give feedback.
-
|
If it's just one word, I would mash it up into a big 10 second file of similar utterances, or try increasing them in intensity throughout the file. Too much odd sounds may cause unpredictable results. You may try a file of all the related grunts. I'm currently working on a model and have one of just the person breathing heavily (for musical effect). The result was that panting, sighing, and related utterances were better understood by the model. You really never know until you try. |
Beta Was this translation helpful? Give feedback.
One second of the speaking part is sufficient.