The dictionary is a file provided with a voicebank that converts words into usable symbols by the voicebank. There are 2 formats to create a dictionary file, with both being equally usable. It is important to see that each method has a different file name format.
Using a Dictionary[]
The dictionary file, for both methods, must be placed in the singer folder with the rest of your voicebank's files. Both formats have a left hand side and a right hand side (to the left or right of the : or =). The left hand side will be the words you can enter into deepvocal. The right hand side is what symbols deepvocal will turn the word into. For words with more than one symbol, deepvocal will take the note and split it into evenly sized chunks to fit the symbols into. This means you will have to time everything manually.
Reminder that if the sound does not play or the system does not read the symbol, make sure that the sounds that are in the dictionary do indeed exist in your library.
 |
 |
|---|
Format 1[]
This is a slightly newer method than the other one. It was created due to some difficulties users had with the second method.
File Creation[]
To start, you must make a utf-8 text document named [WhateverYouWant]-dict.txt. This file needs to encoded in utf-8. This file needs to be placed with the rest of your vocebank file. It can be moved over after editing the dictionary to make things easier.
Word Input and Format[]
To start, the first line in the dictionary must be blank. After that, you can start entering words to phonemes pairs in the following format:[word=Symbol1 Symbol2 ...]. You can also include other words in the pairs in the format of [word2=word1 Symbol1 Symbol2 ...]. It does not seem to like if you place the [word1] anywhere except the start.
Format 2[]
This is the original method which is still commonly used in many voicebanks.
File Creation[]
The dictionary must be prepared in a .txt document and must have the name of your singer followed by the ending .dict, it is important that the name of your singer is the same as the one placed in DeepVocal Tool, as shown in the example :
  |
 |
|---|
The syllable on the left side is the word you will enter into DeepVocal and the one on the right side is the sound that the software will play.
Word input and Format[]
For the dictionary to be readable by DeepVocal, the first line must be a { and the last line a }. To enter words, the word must be surrounded by ", then followed by a [:], then have the symbols surrounded ["], and finally, if there is a word after you must but a [,]. The last word however does not need a [,]. An example using vocaloid phonemes is: ["bya ":" b'a ",].
