Two ways to decide a voice
The first is to choose from voices published on Fish Audio. In the character editor, use "Fetch from Fish Audio" next to "Voice preset" to search by voice name, listen to samples and import one. Importing costs no coins. If the publisher makes the voice private or deletes it, that voice can no longer be used, and the default voice reads in its place until you choose again.
The second is to register a voice from a voice sample. From "Voices" in the left menu, open "New voice" and upload an mp3, wav, m4a or ogg audio file (up to 20MB), and the AI learns that voice. About 10 to 30 seconds of clear speech is recommended, and adding a transcript of exactly what is said improves accuracy. Registering uses coins.
You may only register your own voice, or audio whose owner has given you permission to use it, as a voice sample. Using the voices of voice actors or celebrities, or audio clipped from anime or games, without permission may infringe the rights of those people or the rights holders.
Set a voice on a character and preview it
On the Overview tab of the character editor, choose a voice from "Voice preset". In the same place, you also adjust the pronunciation and speed.
- Speech speed: 1.00 is normal. If left empty, the voice preset's speed is used. The speed of anime lines is set line by line
- Name pronunciation: set this when the voice mispronounces the name. It is replaced only when read aloud, and the displayed name does not change
- Other pronunciations: register up to 20 easily mispronounced words, such as skill names and place names. The name pronunciation and these pronunciations are used in anime lines, chat and story read-aloud alike
- "Play preview": listen to the character's voice with the current settings to check it. Previewing costs no coins
Where the voice is used
The voice you set on a character is used in the following places. Even for characters already published, saving the voice settings reflects them on the side of the crom manga app.
- Anime lines: when you make an anime from a script, the audio for that character's lines is created in their voice. No audio is created for lines of characters without a voice, so decide the voice first
- Character chat: for characters with a voice, readers can listen to messages as audio in crom's chat
- Story read-aloud: when readers enjoy a story read aloud in crom, the lines are read in the character's own voice
- Multilingual dubs: lines in the languages you made anime subtitles for can be synthesized in the same voice to make a dub
Decide the narrator voice for narrative text per work
The voice that reads anime narration and the narrative text (action lines) of stories is decided per work, not per character. Choose a work under "Narration" in the left menu, set the narration voice, reading speed and reading dictionary, and preview it. Until you set the narration voice, no audio is created for narration lines in anime.
Narration for promo videos is not linked to a work, so register it as a "promo narrator" separately from the work's narration.
Listen to anime lines one at a time and confirm them
The audio for anime lines is created automatically right after you make an anime from a script. The lines are listed by scene, so listen to them one at a time with "Preview" and press "Confirm this line" if it sounds right. Once all lines are confirmed, you can move on to production preparation. Creating, previewing and confirming line audio costs no coins.
The audio for a line uses not only the name pronunciation and reading dictionary of the character speaking, but also the name pronunciations of the other characters who appear in the same anime. Narration lines use the work's narration dictionary and the characters' name pronunciations. The dictionaries are applied when the audio is created, so after you fix a dictionary, recreate the lines that are not yet confirmed to reflect the change. If a line still bothers you, fix it on that line alone.
- Change the reading: replaces only the text that is read aloud, while the subtitle text stays as it is. Dictionaries are not applied to a line where this field is filled in
- Voice, emotion and speed: for each line you can choose a different voice, add an emotion and change the speed. After changing them, recreate the audio with "Save and generate the audio"
- Mouth movement: choose "Move" for lines spoken on screen, and "Do not move" for inner voices, narration and off-screen voices
- The voice is out of sync with the mouth in the video: if the voice position does not match after the video is made, you can move just that line's voice with "Earlier" or "Later"
Multilingual dubs
When you bring your anime to audiences overseas, you can prepare subtitles and dubs. A dub is an audio track that synthesizes the lines of a language with finished subtitles in the same voice as the character. The video is not remade; the dub becomes selectable as audio at the time of distribution.
Press "Prepare subtitles and dubs" on the anime screen, and the system decides which of translation, re-joining the finished video and dubbing is needed and runs them in order. Mouth movements are made to match the original language, so mouth shapes do not match in other languages. Creating subtitles and dubs uses coins, and the coins needed are shown on screen before you press it.
